Policy makers and business leaders have developed an urgent interest in understanding supply chains for business disruption, resilience and planning needs. But, finding good data on supply chains is difficult. Fortunately, there may be one untapped data source that enables us to better understand and reconstruct supply chains – payments data.
Traditionally, to analyse supply chains, we have relied on input-output tables that link economic sectors within and between countries or commercial data vendors.
The sector level datasets can help answer macroeconomic questions, such as how a sector X in country Y would be affected by a coronavirus-related shutdown in country Z. Yet, the data also have two major limitations. First, it is rather outdated. Most Input-Output tables I reviewed date back to 2014 or 2016. This makes the data ill-suited for any real-time monitoring or early-warning applications. Second, the sectoral-level of aggregation might not be granular enough. At the moment, we are interested in the supply chains of ventilators or face masks. Or, in a failure of a particular big firm and its effects. Unfortunately, these are all questions which sector-level data cannot fully answer.
Then there are datasets on company relationships from commercial data vendors. These aggregate company fillings, regulatory reporting and press releases to identify major suppliers of major corporations. However, this data is also not real time and miss possibly important smaller linkages that are not published.
This is where a wider use of granular payments level data for a supply chains reconstruction could come into play. It can serve as a basis for analytical models uncovering supplier-buyer relationships. If a firm A makes a payment to firm B, then firm B is a supplier to firm A. The Figure below shows an example of a filtered supply chain recovered from payments data. The direction of the arrow indicates that a firm is a buyer of a good supplied by the other firm.
Stylized example of Supply chains relationships reconstructed from payments data. Node size corresponds to node’s importance in the ecosystem and colours represent the country of domicile of the firm (Yellow for China, Blue US, and Orange South Korea) – FNA R&D Labs
The advantage of using payments data over other datasets to reconstruct supply chains is twofold. First, payments are more granular and display company-level linkages instead of sector-level linkages. They are also more timely, allowing to set-up early-warning alerts to detect disruptions. Supply chains from payments data can help us watch the world unfold in real-time. In times of COVID-19 this could help decision makers identify changes in the economy resulting – for example – from a lockdown of a region. For example, in the Figure above we see that Apple is dependent on Hon Hai for components and manufacturing, also known as Foxconn. If we see order cancellations due to weak demand for Apple products (as suggested here), this could have second-round effects on Foxconn’s suppliers such as Samsung, which would not receive any orders either. Similarly, lockdown impacts at Foxconn would affect Apple’s ability to source devices.
So, where can one access payment-level data? There are a number of different types of organisations with good access. There are large global banks which operate their own internal payment systems in many cases as complex as national interbank payment systems. Also payment systems and payments data repositories keep records of payments made for regulatory purposes.
The current crisis clearly highlighted the benefits of increasing the transparency of supply chains – indicating that it should be high on the policy makers’ agenda. Payments data can play an instrumental role in this effort – the tools to operationalise it are already available.
Lubos Pernis (firstname.lastname@example.org) is a Data Scientist at FNA.