Which Factors Matter in Location Patterns of Restaurants and Bars? A Longitudinal Analysis of Temporal, Ethno-Demographic, Socio-Economic and Accessibility Attributes

We develop a novel dynamic longitudinal Poisson model that can handle complex temporal shifts and identify the effects of ethno-demographic, socio-economic and connectivity/accessibility attributes in location patterns of bars and restaurants.Two key gaps are observed in the extant literature -not accounting sufficiently for any complex temporal patterns and for the urban or rural area characteristics. We fill these gaps by using a unique dataset of two carefully selected UK city-regions - 626 small areas in Greater Manchester and 658 small areas in Nottingham over a 17-year period (2002-2019). These study areas offer varying scales of urbanity and rurality that play fundamental role in location choice decisions. The results highlight complex temporal patterns as reflected by the non-linearities in time fixed-effects and show contrasting increases in restaurants (39.80% and 36.08% increases in Manchester and Nottingham) compared to reductions in bars (22.69% and 23.13% reductions in Manchester and Nottingham). We only find bars to be positively affected by increased retail activity (by 0.34% in Manchester and 0.49% in Nottingham). While broad ethnic categories, such as Black and Asian, show effects in our models, the recognition of ethnic sub-groups substantially changes the results and offer more nuanced understanding.