Description:
ABSTRACTAgglomeration economies explain why productivity increases as the size of agglomeration increases, and thereby partly explain the importance of location and space in economic development. Agglomeration economies are based on the distance between firms, labour and the market. Infrastructure investment can be a tool used by governments to affect this distance, where, in doing so, they induce agglomeration economies and increase economic growth. In this paper, we will discuss the importance of considering types of networks when analysing agglomeration economies using multiregional new economic geography (NEG)‐based spatial general equilibrium models. These models typically use abstract bi‐regional origin‐destination networks to describe the cost of transporting or commuting between regions. These abstract networks are like tunnels between regions and do not take into account the actual roads in real infrastructure networks used for transport and traffic between destinations. In other words, in real infrastructure networks, which we label full networks, people and goods move from one region to another via actual roads and possibly via third regions. In the empirical economic literature, there is a significant difference between the economic effect of changes in a bi‐regional origin‐destination network and the economic effect of changes in a full network. To date, no systematic explanation has been given for this difference in the size of the effects. Lack of an explanation for this difference may lead to confusion among policy‐makers who have to decide on infrastructure investment. In this methodological paper, we discuss the reason for this difference. We do not discuss the absolute size of the agglomeration effects, only the relative difference that exists when using different types of networks. In order to explain the difference, we introduce a methodology that translates the effects in a bi‐regional origin‐destination network into similar effects in a full network, making it possible to systematically compare the economic importance of all links (roads) in such a network. This is highly relevant to policy‐makers, who can now interpret the model results on a real full infrastructure network. We will also show the mathematical relationship between the importance of agglomeration economies on a link in a full network and the importance of agglomeration economies on a link in a bi‐regional origin‐destination network. Finally, the methodology is illustrated with a case study from the Netherlands.