Policy Design from a Network Perspective: Targeting a Sector, Cascade of Links, Network Resilience

Drawing on input-output data, a computational methodology is proposed to: (i) characterize the upstream and/or downstream network of a targeted (or prioritized) sector i, (ii) uncover the cascade of layers of links in the network constructed, and (iii) measure the degree of network resilience using edge betweenness centrality measure of edges between communities. These objectives are accomplished through three complementary algorithms. The implementation of the algorithms is illustrated using Turkiye’s 2018 input-output production network. Ways to design policies are discussed from a network perspective. The key findings are three-fold. First, in network-based policy design, it is highly critical to consider the interdependencies of regulated and seemingly competitive sectors. Efficiencies gained in liberalized markets via pro-competitive PMR can easily be wasted before final consumers benefit from them as regulated industries may exercise their market power to confiscate part of the efficiency gain created in competitive markets. Improved competition in a single market may not generate the desired outcome even if competition policies perfectly support that market because benefits from competition may not spread over the rest of the network due to disruptions in the cascade of interdependencies concerned. Second, a network-based policy design should start with the identification of the “dominant” source and the “subordinate” sink sector(s), and those in between. The source−sink structure of Turkiye’s manufacturing network illustrates that the manufacturing sector is the most dominant, whereas telecommunications and transport, energy and construction sectors are the potential sinks where large chunk of input flow ends up. Agriculture, finance and oil extraction-mining seem to be interactive sectors. Third, the cascade of three layers of links are identified, and the upstream network of the manufacturing sector is found to have a mediocre level of resilience against the complete disruption of the intermediate layer of the network.