“How would an armed conflict, an epidemic, or a flood impact the economy? The ability to assess—and potentially predict—the economic repercussions of such crises is crucial for mitigating and counteracting the damage,” remarks Christian Diem from the Complexity Science Hub (CSH).

A groundbreaking study in PNAS Nexus has unveiled a new frontier in crisis preparedness. It emphasises that countries could significantly enhance their readiness by harnessing an exceptionally detailed supply chain data set. For the first time, CSH researchers have demonstrated that the conventional sector-level economic data often underestimates the economic impact of crises by as much as 37%, in comparison to highly detailed company-level data.

This unique study utilised a comprehensive dataset of value-added tax (VAT) information from Hungary, encompassing 243,399 companies and over 1.1 million supply relationships, representing the entire national economy. Stefan Thurner from the CSH explains that this extensive dataset enabled the researchers to conduct a thorough analysis comparing the economic effects of crises, using either the 88 economic sectors defined by the European Union or detailed supply chain data at the company level, which includes all companies and their customer-supplier relationships.

The research involved simulating 1,000 hypothetical crisis scenarios to ensure that the models closely mirrored real-world crises. The scenarios were informed by empirical data on the economic effects of the Covid-19 crisis in early 2020.

It was a significant finding that the impact of each simulated crisis was consistently underestimated—by up to 37%—when only sector-level data was used, which has been the traditional method. Conversely, the results at the company level aligned much more closely with the actual recession outcomes in the second quarter of 2020, indicating that sectoral-level assessments typically underestimate the full extent of a crisis’s impact compared to company-level data.

Thurner further elaborates, “Traditionally, a country’s economy has been viewed predominantly through the lens of entire economic sectors. For instance, discussing the overall impact on the automotive industry due to supply bottlenecks. However, this new dataset allows us to observe the ‘atoms’ of the economy—the individual companies—and how they interact within the supply chain. This offers a novel and fascinating perspective into economic science.”

The approach allows for a nuanced calculation of how individual companies within sectors are affected by a crisis instead of generalising across an entire industry. “The difference is substantial between stating that a sector might incur a 20% loss and using simulations to identify which specific companies within that sector are likely to be affected,” Diem adds. This method also highlights how these effects propagate through the supply network, impacting direct and indirect trading partners.

Globally, more than 160 countries implement a VAT system and could theoretically use it to reconstruct their supply networks. However, only a select group of countries, including some EU members like Spain, Belgium, and Hungary, as well as nations such as India, China, and some in Africa and Latin America, actively collect the necessary data to facilitate this kind of analysis.

For nations like Germany and Austria that currently do not collect VAT data in a manner conducive to this analysis, a minor adjustment in the VAT reporting by firms could bridge the gap, potentially automated through firm accounting software, according to Diem. This adjustment could not only aid in reducing VAT fraud but also significantly contribute to their crisis preparedness.

Diem concludes, “This study underscores the significant discrepancies between aggregated sector-based estimates and those derived from detailed company-level data, highlighting the importance of collecting and analysing granular data. Accurate data is imperative, irrespective of the nature of the threat—be it a natural disaster, environmental issues, or political interventions—as it enables authorities to anticipate potential consequences and respond swiftly and effectively.”

More information: Christian Diem et al, Estimating the loss of economic predictability from aggregating firm-level production networks, PNAS Nexus. DOI: 10.1093/pnasnexus/pgae064

Journal information: PNAS Nexus Provided by Complexity Science Hub