Despite the widespread implementation of road travel restrictions in Iran during the COVID‑19 pandemic, there is limited quantitative evidence on the spatial effectiveness of these policies at the intra‑provincial scale. This study aimed to examine the impact of travel restrictions on the spatiotemporal dynamics of COVID‑19 in Kurdistan Province and to decompose their direct and spillover effects. To this end, a balanced panel of the 10 counties of Kurdistan Province was constructed for the period 2019–2022, and spatial autocorrelation was assessed using the global Moran’s I statistic. A random‑effects spatial autoregressive (SAR) panel model was estimated with a K‑nearest neighbor spatial weight matrix. The policy variable—the proportion of months under inter‑city travel bans—was derived from the official decrees of Iran’s National COVID‑19 Taskforce and entered into the model through interaction terms with traffic and migration variables. Direct, indirect, and total effects were computed via the inverse of the (I – λW)⁻¹ matrix, and the robustness of the results was examined using six alternative spatial weight matrices. The global Moran’s I statistic was non‑significant in all years (values near zero, *p* > 0.05); however, the SAR model uncovered a strong and significant spatial autocorrelation coefficient (λ = 0.812, *p* < 0.001), indicating that univariate tests may conceal spatial dependence structures in small samples. For all variables, spillover (indirect) effects dominated direct effects (e.g., main road density: direct effect = 2.275 vs. indirect = 5.706). A significant positive interaction between the proportion of restriction months and migration (β = 0.271, *p* = 0.045) revealed that, during periods of strict travel bans, essential migration became the dominant transmission channel. These findings remained consistent across all six weight matrices. Blanket, uniform road travel restrictions, in the absence of the management of essential mobility, shifted the pathway of disease transmission from general travel to exempted migration. Furthermore, policy evaluation in areas with few spatial units should not rely solely on simple diagnostic tests but requires the use of advanced spatial models. Containing future epidemics will demand geographically targeted interventions integrated with screening protocols for mobility exempted from general bans.