Production networks form the structural core of our society’s metabolism. The development, production, recycling, and disposing of intermediate and final goods in any economy is typically spread across several hundreds of thousands of companies that are connected through millions of buyer-supplier dependencies.
These production networks (PNWs) are subject to continuous change through the entry and exit of firms. Connections between existing firms are also subject to continuous restructuring, in the form of new buyer-supplier links forming and existing connections being terminated.
In this talk, I statistically quantify the temporal evolution of an empirical PNW by examining the entry and exit of firms as well as the formation and termination of buyer-supplier connections. To this end, I use monthly reported value-added tax data in Hungary in the period from 2014 to 2022 containing 711,248 companies and 38,644,400 connections.
Our results allow us to calibrate a simple network generation model that reproduces the stylized characteristics of the Hungarian PNW, with on average of 18,800 firms and 28,000 links. The model accurately reproduces the in- and out-degree distribution, as well as the assortativity and local clustering structure. It is realistic enough to reproduce the empirical pattern of the Economic Systemic Risk Index (ESRI).
Our work is useful for improving the modeling of rewiring dynamics for shock propagation and can be used to generate a time series of monthly PNW snapshots for researchers who don’t have access to empirical data.