<![CDATA[Tourism is expected to enlarge subnational fiscal space, but observable tourism activity does not automatically translate into local own-source revenue (PAD). This study estimates the associations between tourist arrivals, registered hotels, registered restaurants, and PAD across the ten districts and cities of West Nusa Tenggara, Indonesia, during 2017–2024. The balanced panel comprises 80 district-year observations compiled from official statistical, tourism, and regional-revenue agencies. Pooled, fixed-effects, and random-effects specifications were compared. The Chow test rejected the pooled model, the Hausman test did not reject the random-effects estimator (χ² = 1.622, p = 0.654), and the Breusch–Pagan Lagrange-multiplier test rejected pooled ordinary least squares (χ² = 120.883, p < 0.001); accordingly, the reported specification is a random-effects generalized least-squares model. Tourist arrivals were positively associated with PAD (b = 153.002 thousand IDR per arrival, SE = 32.596, p < 0.001), as were registered restaurants (b = 281,174.4 thousand IDR per unit, SE = 112,616.4, p = 0.015). The hotel count was negative but statistically inconclusive (b = −108,990.1 thousand IDR, SE = 92,628.1, p = 0.243). The model was jointly significant (F = 11.718, p < 0.001) and explained 28.9% of the adjusted variation. Because the source output does not report shock controls, robust covariance estimates, or the row-level replication file, the coefficients are interpreted as conditional associations rather than causal effects. The results favor policies that convert visitor spending into traceable taxable transactions and improve occupancy, business registration, and digital tax administration rather than simply increasing facility counts.]]>
