Croce studies pandemic risks across markets and crisis stages

Mariano Massimiliano Croce, Professor at Bocconi University and research fellow of the Baffi Centre, is also the principal investigator of “Pandemic Risks across Markets and Crisis Stages” research project, which has just been concluded. We have asked him some questions about the project.

Who has funded this research project? 

Ministero Università e Ricerca (MUR) and PRIN financed within the Next Generation EU initiative have financed this project.

Why this project?

Pandemic crises have become a more frequent phenomenon and are predicted to become even more likely in the future. Given these considerations, they represent an important source of risk at both the local and the global level. The distinctive objective of our research project is the examination of economic dynamics in several markets across different stages of a pandemic crisis. The recent literature on the COVID-19 pandemic situation has highlighted very interesting socio-economic dynamics mostly at the onset of the crisis. How markets adjust in the aftermath of an epidemic shock is still an open question.

We intend to focus on a “long-run’ perspective, that is, study the interactions between the real economy and financial markets over an entire pandemic cycle. Given the interconnection between production activities and the distribution and level of households’ income and wealth, we believe that a proper analysis must be broad and granular. That is, we must study several markets and several sectors across several geographical regions. Since pandemic crises represent extreme (tail) events, we intend to produce and share new nonconventional measures of risk. In addition, in order to maximize the social impact of our project, we will focus mainly on European countries.

Why your research is innovative?

Our data are non-conventional for two reasons. First, some of our indicators are related to information embodied in non-conventional financial assets (such as equity futures, options, crypto currencies) very useful to assess tail-events. Second, some of our estimators were obtained from non-conventional sources, i.e., by collecting announcements and news from social media across several locations. We have adopted both new text analysis methods and new econometric techniques for tail events. In addition, we have looked at the interplay between asset prices, announcements and news shocks collected in real time from social media (e.g., Google, Twitter).

Why the topic of the research matters for people? 

This project has two main goals: broad social impact, and strong academic impact. Our data collection offer to the community at large new indicators that can be used to measure economic, social, and financial distress in real time. Our prominent focus is on pandemic risk, but we believe that in the future we will be able to expand our techniques to other tail events. Providing high-quality, filtered, and readily available nonconventional high-frequency indicators will be useful to policy makers, investors, corporations, and other researchers. This is indeed the first metric on which we will assess the success of this project with respect to academic impact.

Which data have you analysed? 

Twitter, as well as other social media platforms or search engines (e.g., Reddit and Google), enabled us to collect in real-time the flow of news on epidemic risk that can inform us on the evolution of both local and global contagion risks. Modern text analysis tools (Machine Learning) process large numbers of news (Big Data) and capture both their tone and diffusion. These variables are important because tone, topic, and speed of diffusion of the tweets can predict economic activity, financial markets reactions, and can help policy makers in designing prompt and more effective policies. Since tweets can be sorted across both geographical and social segments, they can inform us on local and global conditions. Furthermore, we linked them to different members of society, that is, consumers, families, businesses, government agencies. By analyzing the tone of the tweets across both regions and social segments, we offered a broad set of real-time indicators on social sentiment that are useful to predict future economic and financial performance. These indicators are very relevant because they can be applied to other interesting scientific questions that go beyond economic and financial activity. For example, one could predict social distress, economic inequality, regional imbalances, and crime activity. We think that our data effort can produce broad research with very relevant social gains. Similarly, Google searches or activity on social platform like Reddit can be sorted geographically, providing further information on local and global conditions, sentiment, and economic indicators.

Which were the research questions?

We intend to focus on a “long-run’ perspective, that is, study the interactions between the real economy and financial markets over an entire pandemic cycle. Understanding both resources reallocation and prices adjustment during and after the current crisis will be extremely useful to inform households and policy makers on the best course of action during future episodes. Given the interconnection between production activities, corporate financing and the distribution and level of households’ income and wealth, we believe that a proper analysis must be broad and granular, i.e., it must study several markets and several sectors across several geographical regions. We organize our research agenda in two parts: the measurement side of pandemic risks: new high-frequency indicators and financial markets; market implied risk measures and macroeconomic forecasts.

What are the main findings of the research project? 

The project explicitly aimed to (i) develop granular real-time indicators, (ii) study funding and segmentation frictions across markets, and (iii) assess their predictive and policy relevance. 

These objectives have been achieved through the development of a comprehensive empirical and theoretical framework linking cryptocurrency market frictions to fiat currency funding conditions. 

A novel and non-conventional high-frequency dataset has been constructed combining:

• Daily Bitcoin prices across 13 fiat currency pairs (CryptoCompare, cross-checked with alternative sources),

• Daily spot and forward exchange rates,

• Short-term interest rates to compute cross-currency CIP deviations,

• Equity returns at global, regional, and country levels.

From these data, we constructed:

• The Bitcoin discount (law-of-one-price deviations across fiat locations),
• The BTC basis, measuring no-arbitrage deviations in hedged crypto positions, and the Fiat (CIP) basis, following Du et al. (2018),
• A crypto-specific friction wedge, obtained by orthogonalizing BTC basis with respect to the fiat basis, and a crypto friction news shock, defined as the innovation component of the residual wedge (AR(1) corrected).

The crypto frictions news shocks represent new high-frequency indicators of market segmentation and funding stress that are observable in real time and constructed from non-traditional financial assets, fully in line with the project’s objectives. The project documents a strong and systematic co-movement between (i) deviations from covered interest parity in fiat currency markets, and (ii) deviations from no-arbitrage conditions in Bitcoin markets. This finding is new. It shows that crypto and fiat markets are not orthogonal but are jointly shaped by balance-sheet constraints and funding conditions. The evidence is robust across: (i) G6 and non-G6 currencies, (ii) Country-level regressions, and alternative aggregation procedures. The BTC basis amplifies the volatility of fiat funding frictions, indicating that crypto markets act as a magnifier of underlying financial stress.

Predictive Content for Equity Markets. Using a local-projection framework, we show that:

• Positive crypto friction news shocks predict lower subsequent equity returns.

• This predictive power holds at global, regional, and country levels.

• The result survives controls for standard financial predictors (VIX, high-yield spreads, EM spreads, Treasury yields, USD index).

These results demonstrate that the constructed indicators are not only descriptive but economically meaningful. They contain forward-looking information about risk sentiment and expected asset returns. This outcome directly fulfills the objective of producing indicators that can inform policy makers and practitioners about evolving funding conditions and financial stress. 

Structural Interpretation. To rationalize the empirical evidence, we developed a tractable equilibrium model in the spirit of Gabaix and Maggiori (2015), with: (i) Segmented markets; (ii) Constrained intermediaries; and (iii) Costly balance-sheet expansion in both fiat and crypto segments. The model reproduces:

• Persistent basis deviations,
• Joint movements of fiat and BTC wedges,
• Predictability of asset returns following funding shocks.

This model provides structural discipline to the empirical findings and strengthens the scientific contribution.