This paper takes a broader look at how vertical linkages can trigger the spatial agglomeration of economic activity in a 'new economic geography' (NEG) set-up. First, it formally establishes the key positive features of a wide class of vertical-linkage models without resorting to numerical simulations. Second, it proposes an analytically solvable model of this class. Third, it addresses the important though neglected issue of whether in such models market forces yield too much or too little agglomeration. It shows that, in terms of positive implications, vertical-linkage models are identical to migration models once considered in their 'natural' state space. Important differences arise, however, in terms of normative implications in the absence of interregional transfers: in migration models agglomeration is necessarily bad for people stuck in lagging regions; in the vertical-linkage models it can be good for everybody as it delivers richer product variety.