Using merged administrative datasets from Minnesota, we bring new evidence on the labor market effects of large minimum wage increases by examining the policy changes implemented by Minneapolis and Saint Paul. We begin by using synthetic difference-in-differences methods to estimate counterfactual outcomes at the zip code level from Minnesota and at the city level from the rest of the country. The minimum wage did not affect employment in most industries but exerted a negative impact on restaurants' employment, with an elasticity of -0.8. Next, using variation in exposure to the minimum wage across establishments and workers within the Twin Cities, we find employment effects that are half as large as those from the time series. The cross-sectional estimates difference out employment effects from the pandemic or civil unrest that could confound the time series comparisons, but they do not include potential effects of the minimum wage operating through equilibrium adjustments such as entry. We quantify a model of establishment dynamics to reconcile the different estimates and argue that they plausibly reflect lower and upper bounds of employment losses. We use the model to show that our estimates are consistent with an establishment elasticity of labor demand of -1 and illustrate how they can inform deeper parameters characterizing product and labor market competition, factor substitution, and establishment dynamics.