Spatial Agglomeration, Production Technology and the Choice to Make and/or Buy: Empirical Evidence from the Emilia Romagna Machine Tool Industry

Relying on a new firm–level dataset, this paper focuses on the factors underlying both the decision, and the relative intensity, to either fully or partially source production activities by small firms operating in the indus- trial machinery and equipment sector of Emilia Romagna (Italy). Particular attention is devoted to spatial agglomeration variables, as capturing poten- tial proximity effects that may reduce transaction costs and favour vertical disintegration. We estimate a Hurdle Negative Binomial model where the de- pendent variable is the number of product transformation stages contracted out, and the covariates include firm size, age, labour cost per employee, the stages operated by the firm, and three spatial agglomeration variables, mea- suring specialisation vs. diversification economies at the level of local labour system. Our estimates show that, once controlled for endogeneity and for the composition of the product transformation process, the number of fully out- sourced activities is positively related to the share of skilled personnel, the engagement in foreign markets and the density of neighbouring employment in related three-digit industry. Differently, concurrent sourcing decisions are affected by firm size, age, and unit labour cost, whereas no effect seem to emerge with respect to spatial agglomeration variables. The insensitivity of concurrent outsourcing to the spatial concentration of an industry reflects the lower costs of transaction compared to full outsourcing of activities.