In 1956, 52$\%$ of urban men and 42$\%$ of rural men smoked cigarettes. By 2010, the disparity had flipped: 24.7$\%$ of urban men and 30.6$\%$ of rural men smoked. Smoking remains the greatest preventable cause of mortality in the United States, and understanding the underlying causes of place-specific differences in behavior is crucial for policy aimed at reducing regional inequality. Using geocoded data from the National Longitudinal Survey of Youth 1979 Cohort, I estimate a dynamic model that captures smoking behavior, location decisions, and education over thirty years. Simulation of the estimated model demonstrates that selection on permanent unobserved variables that are correlated with smoking cessation, both in native populations and in those who migrate between rural and urban areas, explains 70.9$\%$ of the urban/rural smoking disparity. Alternatively, differential tobacco control policies explain only 9.6$\%$ of the urban/rural smoking disparity, which suggests that equalizing cigarette taxes across regions may fail to bridge gaps in behavior and health. This paper emphasizes that rural smoking disparities are largely driven by who selects into rural communities.

1956年52岁$％$ 城市人口和42$％$农村男人抽烟。到2010年，差距已经翻转：24.7$％$ 城市人口和30.6$％$农村男人抽烟。在美国，吸烟仍然是最大的可预防死亡原因，而了解特定场所行为差异的根本原因对于旨在减少区域不平等的政策至关重要。使用来自1979年全国青年纵向调查的地理编码数据，我估算了一个动态模型，该模型可以捕获三十年来的吸烟行为，位置决策和教育。对估计模型的仿真表明，无论是在本地人口还是在城乡之间迁移的人群中，都选择与吸烟停止相关的永久性未观察变量，解释了70.9。$％$城乡吸烟差异。另外，不同的烟草控制政策只能解释9.6$％$城乡之间的吸烟差距，这表明在地区之间平均分配香烟税可能无法弥合行为和健康方面的差距。本文强调，农村吸烟差异主要由谁选择进入农村社区来驱动。