This study aims to estimate the effect of disability on household income in Thailand using different econometric approaches and to examine heterogeneity across the income distribution. We analyze data from Thailand’s 2021 Socioeconomic Survey covering 46,775 households, of which 4255 (9.1%) report having at least one disabled member. Employing three complementary methods—ordinary least squares regression, propensity score matching, and quantile regression—we find that households with disabled members experience significant income penalties. The OLS estimate with full controls shows a 10.0% income penalty, while propensity score matching yields 17.4%, suggesting that standard regression underestimates the true effect. Quantile regression reveals striking heterogeneity: the disability effect ranges from 4.8% at the 10th percentile to 30.5% at the 90th percentile. This pattern suggests that among Thailand’s poorest households, both disabled and non-disabled families face universal constraints; leaving minimal scope for disability is strongly associated with reduced household income, while at higher income levels disability creates barriers to advancement. Decomposition analysis indicates disability affects income through both reduced hourly wages (8.2% lower) and fewer work hours (12.4% reduction). These findings reveal that Thailand’s disability allowance of 800 baht per month—representing only 29% of the poverty line—is grossly inadequate, covering merely 17% of the observed income gap. The results highlight urgent needs for allowance increases with inflation indexation, differentiated support across the income distribution, improved employment quota enforcement, and streamlined registration to address the 46% unregistered rate. Policymakers should prioritize raising the monthly allowance to a level commensurate with the national poverty line, implementing tiered benefit structures based on disability severity, and strengthening employment quota enforcement mechanisms to reduce disability-related income inequality.
Wattanadumrong et al. (Sat,) studied this question.