Quantifying firm-level economic systemic risk from nation-wide supply networks

Abstract

Crises like COVID-19 exposed the fragility of highly interdependent corporate supply networks and the complex production processes depending on them. However, a quantitative assessment of individual companies’ impact on the networks’ overall production is hitherto non-existent. Based on a unique value added tax dataset, we construct the firm-level production network of an entire country at an unprecedented granularity and present a novel approach for computing the economic systemic risk (ESR) of all firms within the network. We demonstrate that 0.035% of companies have extraordinarily high ESR, impacting about 23% of the national economic production should any of them default. Firm size cannot explain the ESR of individual companies; their position in the production networks matters substantially. A reliable assessment of ESR seems impossible with aggregated data traditionally used in Input-Output Economics. Our findings indicate that ESR of some extremely risky companies can be reduced by introducing supply chain redundancies and changes in the network topology.

Monitoring supply networks from mobile phone data for estimating the systemic risk of an economy

“…It is enlightening to compare the network topology of the so-obtained RSN (blue in Fig. 2 b) with the topologies of the Hungarian supply network (HSN) (orange) (for which the exact topology is known 18 ) and the private communication network between individual people (green) (i.e. not between companies).…”

“…With a reasonable reconstruction of the supply network, RSN, we turn to the quantification of economic systemic risk in the national production network. For this we use the economic systemic risk index (ESRI) as developed in 18 . The underlying principle of the ESRI algorithm is sketched in Fig.…”

“…For firms producing services we assume that all inputs disrupt their production in a linear fashion. For a detailed explanation of the use of generalized Leontief production functions in the ESRI definition, we refer to SI Text 5 and the “Generalized Leontief” scenario in 18 .…”

“…Systemic risk in these networks was first quantified with network measures like betweenness centrality 9 , which were later improved by explicitly incorporating economic default mechanisms and the associated accounting procedures 10 , 11 . Further extensions involved multilayer networks 12 , 13 , overlapping portfolios 14 , 15 , in the context of financial networks, as well as some applications in the real economy 16 , and lately, also in production networks 17 , 18 .…”

“…Using estimations for production functions for these companies makes it possible to obtain the national production network. Using this as an input, firm level systemic risk for all individual companies were computed by using an appropriately designed SR measure, the Economic Systemic Risk Index (ESRI) 18 . It is a network-based measure to estimate the fraction of the total production output (goods and services) of the economy that is affected by a firm’s (short-term) failure.…”