So far, I've been using the correlation between two industries' location coefficients and alternatively between their absolute regional employment. But there might be more elaborated approaches.
Co-location of firms in space is an agglomeration effect caused by potential increase of productivity via spillovers. There are many literature on that. Like those I take from unpublished paper: 1) Ciccone A, Hall RE (1996). Productivity and the density of economic activity. American Economic Review 86: 54–70; 2) Coll-Martínez E, Moreno-Monroy AI, Arauzo-Carod JM (2016) Agglomeration of creative industries: an Intra-metropolitan analysis for Barcelona. Working Paper Department of Economics and CREIP, Rovira and Virgili University, Nº 16-2016.
I suppose that you have GIS data with firm types and employment and want to introduce some coefficient measuring their relative location in space.You can probably count number of neighbors in neighborhood of certain radius (like done in 2nd paper). You can also weight the neighbor by the number of employee.
In some cases they can be rivals (like in Hotelling model) and would like to stay apart. Different industries can sometimes benefit from synergies, but if they locate closely, they compete for workers. At the same time, they can choose among larger pool of skills. So the issue is rather complicated and probably depends on the exact question under study.
Thanks for the answers. Well, I was just looking for one particular empirical measure, not theory. Currently, I am using the correlation of two industries' location coefficients.