What do financial markets tell us about banking sector risks?

April 6, 2020

Giacomo Tizzanini, Lea Zicchino

A systemic risk analysis based on equity prices confirms that the current crisis can heavily increase banking sector distress through spillover effects that amplify those coming from direct exposure to other sectors. Therefore, new measures are needed to mitigate the credit risk exposure of the banking sector so that it can continue to supply loans to households and corporates


The current health emergency generated by the spread of COVID-19 is causing financial market distress – equity indices have collapsed by about 30% since the beginning of 2020 and corporate bond spreads have increased more sharply than ever experienced. A global recession of unknown magnitude is therefore a certainty.

Although this crisis is different from 2008 as it has not been triggered by a collapse in confidence in the financial sector, at least for now, banks could face heavy costs due to the vulnerability of households and corporates. Defaults could in fact increase sharply, despite the support measures being put in place by governments. In addition, banks could face losses in value on their securities portfolios, at least for the riskiest components, as well as a strong reduction in net interest income and other revenues. On the positive side, banks should be better placed to absorb the impact of the crisis, thanks to the efforts made to strengthen their capital positions and reduce liquidity risks.


This systemic risk measure is directional (as a result of conditioning), because it captures tail-dependency between two institutions, disentangling the interactions occurring on “tranquil times”, in a noncausal sense (these are not Granger-Causality assumptions).

There are several alternative approaches to estimate CoVaR. In our analysis we rely on Multivariate Dynamic Conditional Correlation GARCH (DCC-GARCH, Engle 2002) with normal distribution, instead of quantile regression (Bassett e Koenker 1976), because it needs systematic state variables (macrofinancial indicators) to capture time-variation in the joint distribution.
Following Adrian e Brunnermeier (2016)’s framework, we calculated systemic risk using average weekly returns from January 1980 (source Refinitiv) for all the 18 “super sectors” classified by Borsa Italiana (we expressed results as monthly averages to reduce the noise of higher frequency data).

Figure 1 shows the banking sector riskiness conditional on the distress of each other sector. We can infer that the baking sector is broadly exposed to all sectors, but in particular to Financials, Insurance, Construction and Autos. Figure 2 shows the time series of Banks’ systemic risk: it has not yet reached its peak (Brexit referendum in 2016) but it has experienced a sudden drop not seen since the mid ’90s.
The market based systemic risk measure that we estimated highlights the strong impact that this crisis is having on the banking sector and the need to mitigate its exposure to credit risk so that it can continue to perform its role of extending loans to households and corporates.

Figure 1: Italian Banking sector ∆CoVaR(99%) at March 2020
Figure 2: 5-95% cross-section interquartile range of Italian banking sector ∆CoVaR(99%) conditional on the distress of each other sector.
[1] The ∆CoVaR is strictly related both to tail risks and volatility models (among the others the CaViaR developed by Engle and Manganelli (2004), the systemic risk indicator of Brownlees and Engle (2015)) and the strand of literature on contagion and volatility spillovers (Claessens e Forbes (2001)).
[2] Si veda Adrian e Brunnermeier (2016) per dettagli sul metodo di stima.



Adrian, T. and Brunnermeier, M. K. (2016). “CoVaR”, The American Economic Review, 106(7), 1705.

Bassett, G. W. and Koenker, R. (1976). “Regression Quantiles”, Econometrica, 46(1): 33–50.Brownlees, C. and Engle, R. F. (2017). “SRISK: A conditional capital shortfall measure of systemic risk”, The Review of Financial Studies, 30(1), 48-79.
Claessens, S. and Forbes, K. (2001). “International Financial Contagion”, Springer: New York.
Engle, R. (2002). “Dynamic conditional correlation: A simple class of multivariate generalized autoregressive conditional heteroskedasticity models”, Journal of Business & Economic Statistics, 20(3), 339-350.
Engle, R. F. and Manganelli, S. (2004). “CAViaR: Conditional Autoregressive Value at Risk by Regression Quantiles”, Journal of Business and Economic Statistics, 23(4).

Read All