Secrets of big data supporting companies in distress
26 June 2019
Early Waring machine learning tool gives the opportunity to use data to see issues that could create trouble to SMEs. It is the method that helps identify companies in crisis in order to offer them valuable advice and support.
Time is a crucial factor in a potential turnaround of a suffering company, and relevant advice in the early stages of a crisis, may reduce personal, business-related and socio-economic consequences of the distress. Hence, the 15 Early Warning Europe partners considered different ways to identity or to connect with companies as early as possible.
The ever-growing volumes of public data, combined with EU requirements for standard data structure for accounting across EU Member States, have led the Danish Business Authority, one of the 15 partners, to use machine learning to test the possibilities to exploit publicly available data to distinguish companies in distress from financially well-functioning companies. Machine learning gives the opportunity to use a large volume of public data, to see issues that could create trouble to SMEs.
The tool was tested in five countries in the shape of a program based on Danish data with four derived, tailormade models applying to Poland, Italy, Greece and Spain. It analyses accountancy data from large numbers of enterprises and identifies the probability of distress in a given company. We see a promising start to the use of machine learning and artificial intelligence in helping companies in distress across Europe.
Sometimes there are differences in the quality of publicly available data across the project's participating countries, given the fact that public digitalization is still in an early or transitional phase. For our project this has meant that derived models for the project's four target countries are less accurate than the original base model, but still fairly accurate.
As part of this implementation process, the Early Warning project staff is required to research and confirm the indications prior to any approach, be it physical or other. The output expresses a likelihood or probability and not a prediction. Bear in mind, that other non-financials factors may be at the root of company distress.
“When the project finishes, we expect that this machine learning tool would have helped entrepreneurs not only in 5 different countries, but also validated and implemented in other countries as well”, states Morten Møller, the project manager of Early Warning Europe.
As part of this European effort to support companies in distress, the project has trained personnel to handle company owners in a difficult and complex situations involving personal and financial aspects. Also, it intends for all further use of this and other tools developed within the project to rely on appropriate competencies.
“At the end, we expect that in some years entrepreneurs from all EU countries will have the possibility to be contacted, advised and supported to convert crisis into growth and this contact could be proactive, even before they realise or admit themselves that their enterprise is susceptible to have financial problems. This is just the first step”, finalizes Morten Møller.
More information on Machine Learning tool is available is this concept paper.
You can also find it here