by Philip Fliers (Queen’s University Belfast), Chris Colvin (Queen’s University Belfast), and Abe de Jong (Monash University).
This blog is part of our EHS 2020 Annual Conference Blog Series.
Business bankruptcies are rare events. But when they occur, they can prove catastrophic. Employees lose their jobs, shareholders lose their savings and loyal customers lose their trusted suppliers.
Essentially, bankruptcies are ‘black swan’ events in that they come as a surprise, have a major impact and are often inappropriately rationalised after the fact with the benefit of hindsight. While they may be extreme outliers, they are also extremely costly for those affected.
Because bankruptcies are so rare, they are very hard to study. This makes it difficult to understand the causes of bankruptcies, and to develop useful early warning systems.
What are the risk factors for which shareholders should watch out when evaluating their investments, or when pension regulators audit the future sustainability of workplace pension schemes?
Our solution is to exploit the historical record. We collect a dataset of all bankruptcies of publicly listed corporations that occurred in the Netherlands over the past 100 years. And we look to see what we can learn from taking this long-run perspective.
In particular, we are interested in seeing whether these bankruptcies had common features. Are firms that are about to go out of business systematically different in terms of their financial performance, corporate financing or governance structures than those that are healthy and successful?
Our surprising result is that the features of bankrupt corporations vary considerably across the twentieth century.
During the 1920s and 1930s, small and risky firms were more likely to go bankrupt. In the wake of the Second World War, firms that did not pay dividends to their shareholders were more likely to fail. And since the 1980s, failure probabilities have been highest for over-leveraged firms.
Why does all this matter? What can we learn from our historical approach?
On first glance, it looks like we can’t learn anything; the drivers of corporate bankruptcies appear to change quite significantly across our economic past.
But we argue that this finding is itself a lesson from history.
The development of early warning failure systems needs to take account of context and allow for a healthy degree of flexibility.
What does this mean in practice?
Well, regulators and other policy-makers should not solely rely on ad hoc statistical models using recent data. Rather, they should combine these statistical approaches with common sense narrative analytics that incorporate the possibility of compensating mechanisms.
There are clearly different ways in which businesses can go bankrupt. Taking a very recent perspective ignores many alternative routes to business failure. Broadening our scope has permitted us to identify factors that can lead to business instability, but also how these factors can be mitigated.