Grocery food sales taxes (or grocery taxes) in the United States are applied in the form of a state and/or county tax. To investigate how local governments establish these grocery taxes, we develop a dynamic gaming model to explain the county–county and county–state interactions regarding grocery taxes. Leveraging novel panel data on grocery taxes at county and state levels from 2006 to 2017, we estimate a dynamic spatial model including multilevel governments. The empirical evidence unveils three key spatial determinants that contribute to variations in county grocery tax rates. (1) A negative vertical impact from the home state, (2) a positive horizontal effect from neighboring counties, and (3) a positive diagonal effect from neighboring states.

美国的杂货食品销售税（或杂货税）以州和/或县税的形式征收。为了调查地方政府如何制定这些杂货税，我们开发了一个动态博弈模型来解释县与县之间以及县与州之间关于杂货税的相互作用。利用 2006 年至 2017 年县和州各级杂货税的新颖面板数据，我们估计了一个包括多级政府的动态空间模型。经验证据揭示了导致县杂货税率变化的三个关键空间决定因素。(1) 来自本州的负面垂直影响，(2) 来自邻近县的正面水平影响，以及 (3) 来自邻近州的正面对角效应。