CSH researcher Tobias Reisch will give this virtual talk (research together with Georg Heiler, Christian Diem and Stefan Thurner) at the Networks 2021 conference.
Abstract:
The underlying structure of an economy is the network of customer-supplier relationships. Economic shocks are propagated as demand-side shocks from buyers to suppliers or as supply-side shocks from supplier to buyer. This insight dates back to ideas by Wassily Leontief in the 1930s, who developed modern input-output analysis, which analyses an aggregated version of the customer-supplier network. Mostly due to a lack of appropriately granular data, input-output analysis is typically focussing on sector level economic shocks and shock propagation. Only recently has data on firm-level trade relationships become accessible – restricted, however, to a small number of countries, such as Japan, Belgium, Brasil and Hungary. Typically this data is hard to obtain through extensive surveys, tax or payment system data.
Here we present telecommunication data as an inexpensive and alternative way to get a good estimate of the national customer-supplier network of an entire nation that can also capture international links of firms, a novelty in research. Through cooperation with an mobile phone operator, we are able to analyse mobile phone communication between anonymized companies in a small European country. We compare the communication links with a survey on critical suppliers conducted in April 2020. From this we calculate the probability of correctly identifying a customer-supplier relationship given a communication link. The probability rises with the intensity of the communication to values above 90%.
Using additional external information we estimate quantity and direction of the flow of goods and calculate the systemic risk of individual firms. We calculate a Systemic Risk Indicator for a firm as the fraction of the economy that is potentially subject to cascading effects subsequent to the default of said firm. We investigate the limitations and effects of our assumptions by simulations and outline future research directions.