Urban neighborhoods differ not only in how many restaurants and shops they host, but also in the quality of those firms. We ask how spatial demand, competition, and spillovers shape the distribution of heterogeneous retail firms within and across locations, and how that distribution and consumer welfare respond to work-from-home adoption. Combining mobile-device GPS data with San Francisco administrative records, we document two patterns: firm-level visits vary more within Census block groups than across them, and consumers trip-chain, generating spillovers between neighbors. We then build and estimate a spatial equilibrium model in which consumers make nested location–firm choices and heterogeneous firms enter under incomplete information. Simulating work-from-home adoption, we find that city-center locations gain entrants despite losing foot traffic from local workers. Lower rents as workers leave, combined with these locations’ higher estimated inherent quality, outweigh the lost demand. In locations where entry falls, the welfare loss is smaller than counts alone suggest because the marginal firms are those with the lowest expected visits. Aggregate consumer surplus rises 2.7 to 8.0 percent across adoption levels, driven mainly by better access to retail.

City governments face a trade-off in managing curb space: providing parking to facilitate access to consumption amenities and generate revenue, versus allocating it to alternative land uses. In this paper, we quantify the welfare implications of curbside parking and evaluate alternative policies for managing curb space through parking instruments. We develop a structural model of drivers' joint destination and parking decisions: drivers choose which destination to visit under imperfect information about parking availability, then decide where to park near the chosen destination. We estimate the model using high-frequency data on metered parking transactions and GPS data on visits to points of interest in San Francisco, one of the few cities that have implemented demand-based pricing for curbside parking. We find that, while drivers value curbside parking, the present discounted value of parking revenue and driver surplus generally falls short of local assessed land values, which proxy for the economic value of land uses. Compared to a revenue-maximizing uniform pricing scheme, San Francisco's demand-based pricing generates about 30% more revenue while reducing cruising trips by nearly 70%. Our counterfactuals show that reducing parking supply by roughly 6% and lowering the status quo demand-based prices by $1.25 citywide preserves parking welfare, with only a modest revenue loss, while freeing curb space for other uses.