In the collective imagination of risk, the failure of a company puts both its suppliers and its customers under pressure. Suppliers suffer from a decrease in their turnover, but also from the loss of customer credit. Remember that to compensate for this dry loss, a supplier must generate 10 times the turnover generated by the defaulting company with another customer (assuming a 10% margin). On the other hand, customers have to face a disruption in their supply chain, which can jeopardize their business.

Companies connected to each other through intermediaries

The article "Bankruptcy propagation on a customer-supplier network: An empirical analysis in Japan "This study is based on data from Japanese companies. It proposes a study of the risk of failure of the partners of a failing company. Let's take a quick look at this study and its results.

This one observes the connection of two databases. The first one is the equivalent of the Ellisphere database: a history of legal events and financial information of companies. The second is a database of customer-supplier relationships; it lists contracts between companies and their amounts. Finally, more than one million companies are studied, for more than four million customer-supplier relationships.

Ainif, the analysis of the two databases as a whole shows some interesting relationships. First, there are relatively few direct direct between the companies. Only 0,000338% of the possible links between firms actually exist. Nevertheless, thes ecosystem ofs ecosystem ess veryvery connected since 80% decompanies are connected via one or more intermediaries. The number of intermediaries is even rather low, usually three.

 

The structure of the business network, a key element

Why focus so much on the structure of the business network? Because structure is essential to understand how the shock of a default is borne by the ecosystem. Let's look at a simple case, considering only the financial shock of the default. If the partner company has no other partners, it must bear the loss (of sales and materials) alone. If it has other customers, the loss will represent a smaller share of its turnover, but it could also induce its other customers to reduce their payment terms to rebalance its available cash. Thus, the shock is partly borne by all other customers.

The company can also put its suppliers to work by extending their payment terms, which spreads the impact of the loss over even more companies. We can thus see that commercial links make it possible to mitigate the shock by distributing it among several actors. The real mitigating power of these commercial links remains to be understood, as well as the difference between direct and indirect connectivity.

 

Focus on the Japanese case

In Japan, the network has low direct connectivity (each company has few partners)but strong indirect connectivity. What about the waterfall?

A survival model, a technique used in medicine to evaluate the effectiveness of treatments, was used to estimate the impact that the failure of a supplier or customer could have on the probability of failure of the target company. It was measured that a company tripled its probability of failure if 50% of its customers failed. The impact of a loss of 50% of its suppliers is slightly lower: a factor of 2.

These results indicate an increase in risk, but not in the actual probability of the number of failure cascades. To investigate the latter, the author of the study looked at the direct links between failing firms. It was observed that 90% of failing firms failures do not drag any other company down with them. Of the remaining 10%, an overwhelming majority of cascades involve only two firms. Finally, the author did find some slightly more complex structures. The graph below shows the most complex case involving 35 firms. The arrows represent the supplier->customer direction, and the numbers represent the failure dates with respect to an arbitrary reference.

Cascading defaults, what impact for the economy? 

Cascading default is therefore a significant risk at the corporate level. It represents 10% of insolvencies. But what is the economic impact more globally? The author has carried out numerous simulations for different types of networks, with or without taking into account the measured cascade effect. It appears that cascade failures are very infrequent. The simulations showed that a high degree of connectivity (even with intermediaries) between firms made it possible to cushion the shock generated by the failure of a firm. Each level of intermediary dilutes the loss over more and more firms.

 

What can we remember?

This article on the study of cascading failures of Japanese firms sheds light on many intuitions in the field of risk. In particular, the author shows that the failure of a significant part of a firm's customers and/or suppliers significantly increases (by a factor of 2 to 3) its risk of failure. However, in terms of the ecosystem, the numerous links between firms provide a structure that allows the shock to be spread over a large number of actors and therefore limits the number of firms that could be caught in a cascade of failures.