In containing and mitigating the diffusion of COVID-19, countries are not fully able to pursue local test, trace and isolate strategies due to difficulties in detecting place-based infections and positive asymptomatic cases. This paper explores whether and to what extent local labour markets, functional areas defined by employment self-containment indexes and labour mobility data, can
grasp the spatial dynamics of COVID-19 diffusion. Local labour markets capture most of the socio-economic interactions of working and residential populations and identify areas in which people are more likely to engage in frequent, face-to-face contacts with neighbours, colleagues, friends and relatives. Through an exploratory spatial data analysis and the estimation of a spatial autoregressive model, this paper examined a sample of 441 municipalities and 20 local labour markets. These territorial units belong to the Lombard provinces of Bergamo and Brescia (Italy), among the worst affected areas of the country in respect to both reported deaths and confirmed infections in the early stages of the pandemic. The findings suggest that municipal variations in mortality rates in 2020 correlate with a range of statistics for local labour markets, namely self-containment indexes, labour market dynamics and commuting behaviours. Overall, this paper shows that local labour markets are a useful scale of analysis in detecting the geography of COVID-19 diffusion in the target sample, and verifies the possibility of capturing the spatial dynamics of the epidemic on a smaller territorial scale than NUTS-3 regions do.