Marcellino studies uncertainty and tail risks
Massimiliano Marcellino, director of Baffi Centre and Asset and Risk Management (ASSET) Research Unit Director, is also the principal investigator of “Uncertainty and tail risks” research project. We have asked him some questions about the project.
When have you started this research project?
I have started this research project in 2022, together with a team of 7 Italian researchers.
Who is funding this research project?
Ministero Università e Ricerca (MUR) and PRIN financed within the Next Generation EU initiative are financing this project.
What are the research questions?
The main aims of the “Uncertainty and tail risks” research project are:
- Improving econometric modelling of economic data;
- Providing more robust methods for economic nowcasting and forecasting;
- Developing novel identification schemes to retrieve the impact of macroeconomic and
financial shocks;
- Enriching our understanding of the relationships among economic variables and of the effects of economic policies during crisis time;
- Providing enhanced measures of risk and uncertainty, and of their effects;
- Providing policy advice based on scientifically sound investigations on how to optimally manage the economic impact of tail events, such as the Covid-19 shock.
Why have you decided to investigate this topic?
The outbreak of Covid-19 caused dramatic drops in world GDP and employment, and increases in economic, financial, and political stability risks, partly offset by timely expansionary economic policy. From an econometric point of view, the pandemic represents a tail event, with the war in Ukraine and the Great Financial Crisis as other examples that also caused a massive increase in uncertainty. Hence, for both structural and reduced form analyses, there is the need to develop models that are specifically designed to also handle tail events rather than only the average behaviour of economic variables. These models should also allow for flexible modelling of the error volatility, to permit volatility spikes and persistent clusters of high uncertainty, combined with measures that transform the model outcome into risk measures.
Why this topic matters for people?
Improving econometric modelling of economic data and providing enhanced measures of risk and uncertainty, and of their effects may help policy makers in manage the economic impact of tail events, such as the outbreak of Covid-19.
What are the implications of this topic for policy makers?
Accurate inference on risks that society may be exposed in the future from disruptive events, such as the Covid-19 pandemic, financial crisis and extreme climate events, is of a crucial importance for economic and financial risk managers, policymakers, etc. in order to quantify the losses that local and global economy is subjected to.
Which data are you analysing and which models?
We work mainly with macroeconomic and financial data, exploiting very large datasets with variables sampled at different frequencies. In terms of models, we consider both state of the art econometric models, such as vector autoregressive and factor models, and machine learning models, such as Bayesian additive regression trees, Gaussian processes and neural networks.
What is the state of the project at the moment?
We are revising various articles for publication in academic journals.
Is there any conclusion that you can share regarding your research?
It turns out that large datasets (Big Data) can be particularly informative in problematic periods, such as financial crises and economic recessions. When analyzed with the help of machine learning models, the prediction of tail events and associated risks can be improved with respect to the use of standard econometric models and data.