‘Deserts’ and ‘hotspots’: the evolution of centrality, accessibility, and socio-economic impacts in bar and restaurant location patterns

The literature of bar and restaurant location patterns does not account sufficiently for temporal dynamics and the role of urban spatial. We address these gaps by developing a novel urban economic framework with a two-fold empirical approach: a) dynamic longitudinal Poisson models that incorporate socio-economics, demographics and accessibility to analyse temporal shifts in the concentration of bar and restaurants, or ‘hotspots’; and b) a longitudinal logit approach that models the determinants of bar and restaurant ‘deserts’. We compile a unique dataset of more than 1,100 small areas over a 17-year period (2002-2019) across two carefully selected UK city-regions, Greater Manchester and Nottingham. The key findings reflect a fundamental shift in consumer preference over time. During the study period, the probability of bar ‘desert’ formation increased almost 20-fold along with the reduction in bar counts by around 35 percent in both study-areas. Conversely, restaurant numbers have increased by almost 35 percent and the probability of restaurant ‘deserts’ reduced by at least a factor of 5. While the Poisson specification provides evidence of positive path dependence in areas with an established bar and restaurant ‘ecosystem’, both approaches show significant accessibility, agglomeration, and socioeconomic sorting effects.