Quarrying can contribute to rural cash income, but participation is self-selected and expenditure comparisons do not measure all livelihood costs. This study examined the association between directly reported quarry participation and household consumption using the June 2026 survey in Kinyenze and Lugono villages, Mvomero District, Tanzania. Of 142 household records, 137 contained direct participation responses; the primary analytic sample comprised 86 participants and 50 nonparticipants with observed food expenditure. Reported expenditure components were aggregated and divided by household size; blank non-food components were coded as zero. Augmented inverse-probability weighting (AIPW) combined participation and outcome models adjusting for age, sex, marital status, education, household size, birthplace and village. The adjusted consumption difference was TZS 11,243 per person per month (95% percentile bootstrap confidence interval: TZS 549–20,764; normal-approximation p = 0.022). The food-expenditure estimate was positive, whereas log consumption and the asset/access index were imprecise. Sensitivity point estimates remained positive, but the nonlinear specification’s percentile interval included zero. Residual covariate imbalance, a weighted control effective sample size of 30.3, uncertain expenditure reporting and limited bootstrap replications qualify the findings. AIPW does not establish temporal order or eliminate unmeasured confounding. The results are adjusted cross-sectional associations, not causal effects or evidence of comprehensive welfare improvement. Longitudinal research with documented sampling, complete expenditure measurement and direct assessment of occupational and environmental costs is needed.