Authors:
- Vito Ciciretti, Independent Researcher, Quantitative Risk Methodology Specialist- Berlin, Germany
- Alberto Pallotta, Lecturer, Middlesex university London, UK & head of R&D, D&D Capital Management, UK
- Dr Monomita Nandy, Associate professor in Accounting and Finance, Brunel University London , UK
- Dr PK Senyo, Associate Professor in FinTech and Information Systems, University of Southampton, UK
- Dr Suman Lodh, Associate Professor in Finance, Kingston University London, UK
- Dr Jekaterina Kartasova, Senior Lecturer in Accounting and Finance, Middlesex university London
The sudden shift in the global economy is triggered by the COVID-19 setbacks. As the economic growth of any country is heavily dependent on small and medium sized enterprises (SME) with fewer than 250 employees and annual turnover of less than €50 million, they are worst hit by COVID crisis. Most of the existing academic research papers discussing the impact of the pandemic on SME are focused on limited survey data, which restricts the possibility of applying advanced Machine Learning algorithms to embed the early signal of market wide risk in the SME’s risk modelling. Thus, we are extending our ongoing research on publicly traded companies towards non-traded SMEs.
In the past, we have seen that financial markets experience periods of high volatility in response to sudden changes in systematic risk, for example due to exogeneous events such as the outbreak of COVID-19. For example, the Brexit, the subprime mortgage crisis and the trade war between US-China are only some events worth mentioning. Thus, if there would be a model to capture the early signals of a crisis, investors could timely rebalance their portfolios, risk managers could promptly curtail exposures to systematic risk factors and policymakers could coordinate monetary and fiscal policy changes. Such positive approach of the investors will support the SME as their stakeholders in the supply chain will feel comfortable to carry on their business, there will be no delay in the payment to SME, the demand from the customers will stay stable and overall, there will be a balance in demand and supply of SME products and services.
In a market environment characterised by high systematic risk, investors tend to sell risky securities. This is reflected in the covariance matrix, which is a measure of how much market prices tend to move together in response to market events. However, the investors would be interested to know more about the SME to make them eligible candidate for an investment. Thus, we need a methodology to capture risk signals based on the fundamental data collected from balance sheet, income statement and other financial statements of the SMEs.
We can measure the degree of correlations are expected to change if subjected to exogeneous events – such as the outbreak of COVID-19 – by looking at its eigenvectors and eigenvalues. As covariance matrices change in response to investors’ demand for risky securities, eigenvalues and eigenvectors also respond to investors’ sentiment. Hence, our early-warning risk signalling model is the first and foremost based on observing the eigenvectors and eigenvalues of the covariance matrix of asset returns.
Additionally, we leverage on another way to represent financial markets by means of graph theory. Already employed by social media to represent the interconnections of users, interests and events published on such platforms, we employ graphs to schematize the hierarchical relationships among the securities in financial markets. A similar eigenvector-based measure exists in the case of graphs too, hence we also use the eigenvector centrality of graphs to build early-warning risk signals. In fact, when there is a change in systematic risk, we observe that the graph becomes much denser in response to much more interrelated assets, which mirrors the property of exhibiting higher covariance in higher risk settings. We can apply the graph theory on fundamental data.
The economic crisis generated by COVID-19 is not the last uncertain event for sure. But we learned a good lesson from this crisis and found the need of developing a model to generate early-warning risk signals. In response to this research need, we develop a model based on covariances and graph-theory that is able to provide early-warning risk signals even based on fundamental data of the SMEs. Our research model will build confidence among investors and policy makers and will allow the SMEs to contribute to economic progress with limited disturbance.
If you are interested to know more about the research, please contact the corresponding author at
