The Covid-19 pandemic highlighted the interconnected nature of global supply chains, and showed how a disruption in one part of the world can have global effects. In 2021, supply disruptions were cost the global economy an estimated $1.9 trillion.

In collaboration with Complexity Science Hub Vienna, Oxford INET, and Arizona State University, the Supply Chain AI Lab are calling on government agencies and national banks to support an effort to map the billions of connections in the global supply network which, among other impacts, could reduce tax evasion by as much as €130 billion annually in the European Union.  Understanding supply networks could also improve supply security, promote an objective monitoring of the green transition, and strengthen human rights compliance. Writing in the journal Science, the group discussed a roadmap and emphasised that international alliances, backed by government organisations and the research community, are needed for such an understanding.

Even though most companies know their immediate trading partners, they depend on countless other relationships up and down the supply chain. Understanding supply chain interdependencies between companies, sectors and countries is vital for many challenges, from identifying how disruptions may emerge and cascade across economies, through monitoring carbon emissions and ensuring ethical and sustainable practice. A shortage anywhere in the supply network may affect suppliers, suppliers of suppliers, and so on, as well as customers and their customers.  Supply disruptions caused an estimated loss of 2% of global GDP in 2021 - approximately $1.9 trillion - and significantly contributed to the current high inflation. For a long time, it was unthinkable to analyse the global economy at the company level, let alone its complex network of supply interconnections. That is changing now.

For almost a century, only aggregated data - such as the average values of entire sectors - could be analysed. Predicting how individual company failures would affect the system was simply not possible. What happens to the economy when a specific company stops its production? What if an earthquake paralyses an entire region? Now, a combination of new micro-datasets, methods based in machine learning, and multiple government initiatives are creating the ability to map entire economies, which can give us the tools to answer some fundamental questions with real and timely impact.

Although the volume of data is vast - there is approximately 300 million companies worldwide, each with an average of 40 domestic suppliers, resulting in up to 13 billion supply connections - researchers can map the connections between individual companies.

Currently, value-added tax (VAT) data is the most promising option for reconstructing reliable large-scale supply networks. Countries including Spain, Hungary and Belgium use standardised VAT collection that practically records all domestic business-to-business (b2b) transactions. With these, it's possible to map the entire national trade of a country.  In most countries like Germany, Austria, or France, where VAT is not collected for individual b2b transactions but only accumulated over a specific period, such mapping is not possible. The standardised b2b collection could reduce administrative overheads for companies and would contribute substantially to tax compliance. 

The next step is to link trade data from different countries. Currently, the EU records trade in goods between its member states at the company level. If it also included services and linked them with VAT data, this could lead to comprehensive cross-border company-level network – which would represent almost 20% of the global GDP. 

The European Commission laid the legal foundation by proposing "VAT in the Digital Age." But unfortunately, this is far from being realised. So far, we do not have a single situation where the supply chain networks of any two countries have been joined and merged. This would be an essential next step. To create a truly international picture of supply interconnections, hundreds of datasets must be joined, analytical tools developed, and an institutional framework must be created, together with secure infrastructure for storing and processing enormous amounts of sensitive data.  To advance this endeavour, a strong international alliance of various interest groups is required, including national governments, statistical offices, international organisations, central banks, the private sector, and academia.

The evolution of large-scale firm-level data on supply networks. see Pichler et al.  Science (2023). DOI: 10.1126/science.adi7521

The Supply AI Lab is part of this first interdisciplinary collaboration, aiming to establish a foundation. The researchers hope to inspire others to join their efforts. For more information about this subject, please contact Prof. Alexandra Brintrup.