We propose a method of developing multiscalar industrial policies based on a spatially constrained clustering model which combines aspects of complexity, relatedness and geography. Our formulation is motivated by recent developments of economic complexity regarding the evolution of economic output through interactions among industries within economic regions. This model aims to aggregate a set of geographical areas into a prescribed number of regions such that the resulting regions preserve key interactions among industries and encourage the emergence of complex industries. We use Colombia as a case study to illustrate how our mode can be applied for policy and regional development.
