Around 40% of employees in Canada work within 5 km of their place of residence, highlighting workers' preference for proximity to workplace. I leverage a subway expansion in Vancouver during the 2010 Winter Olympics, and show that workers who gained improved access to the subway network experienced an increase in earnings by 1.5-2%. The effect is driven by job switchers who travel farther to new employers. To interpret these findings, I build, identify, and estimate a two-sided labor market matching model featuring wage-posting, commuting costs, and residential choice. Using the estimated model I replicate the reduced form effects, and show that the expansion reallocated workers to more productive firms. I also show that labor market concentration dropped by 10-35% in treated areas due to the expansion. Separately, I use the model to shed light on the role of preference for proximity in shaping the spatial distribution of wage markdowns.