October 21, 2016

News in Charts: BNP Paribas – The Difference Between a Regional and a Global Banking Crisis

by Fathom Consulting

Using correlation analysis of banks’ common equity prices, we construct a Minimum Spanning Tree showing the most likely path of contagion through the global banking system. Although not necessarily the most vulnerable to a banking crisis, we find that BNP Paribas sits at the heart of the system — meaning that if it were to catch a cold, that cold could spread quickly. In other words, what matters for Europe, matters for the world!


We have long argued that the euro area banking sector is in poor health. Developments this year have confirmed that view. Undercapitalisation, non-performing loans, fines for misconduct and low profitability continue to dog the sector. With the balance sheets of euro area banks already fragile, they can ill-afford another shock, economic or otherwise. But the global financial system is highly connected, and our analysis suggests that the euro area sits at the heart of it. Depending on the size, a shock to any one bank may not be contained within that bank or even the country in which it is based. Instead, the reverberations may be felt far and wide, as the global financial crisis of 2008-2009 taught us. In this week’s News in Charts, we explore how a shock to any one bank may work its way through the global banking system.


Refresh the chart in your browser | Edit chart in Datastream

Mapping the global banking system

We use the correlation between banks’ common equity prices since the beginning of this year as a measure of interconnectedness. Our sample includes 29 of the world’s biggest banks according to balance sheet size, located in the euro area, UK, Switzerland, US, Japan and China. With a correlation coefficient of 0.83, the equity prices of BNP Paribas and Barclays have the strongest relationship year to date.


Refresh the chart in your browser | Edit chart in Datastream

Based on all 406 correlation pairs it is possible to draw a Minimum Spanning Tree (MST) of the global financial system. The algorithm underlying the MST finds the path of least resistance, where resistance decreases with the degree of correlation. Put differently, much like the travelling salesperson who wants to find the shortest possible route on their journey, we want to identify the most direct route through the global banking system. The stronger the correlation between the equity prices of two banks, the closer the connection. By mapping this, the MST (the first chart) reveals the path of least resistance through the global banking system.

In our MST, the size of each node indicates the size of the balance sheet of each bank, with different colours representing different regions. As one would expect, we find clustering within each region since banks tend to be highly interconnected with other banks operating in the same geographical area. The exception is HSBC, which has over 38% of its credit exposure in China. It is, therefore, no surprise that HSBC’s equity price is highly correlated to that of Chinese banks.

BNP Paribas is front and centre…

As our diagram implies, BNP Paribas has the highest mean correlation among the 29 banks within our analysis. As a consequence, it takes centre stage in our map of the global banking system, meaning that if it were to catch a cold, that cold could spread quickly. In other words, once France’s biggest lender is affected, a regional banking crisis could quickly turn into a global one.

…reflecting the diversity of the bank’s credit exposure

Correlation is not causation. However, correlation between banks’ equity prices ought to reflect common risks to the earnings outlook of these institutions. For example, if the business models of two banks have similar shortcomings – Deutsche Bank and Commerzbank spring to mind – the share prices of these banks ought to be correlated. By the same token, the share prices of banks with similar counterparty risks and credit exposures ought to move in line with each other. For the latter, data is readily available. Indeed, as part of its stress test results, the European Banking Authority (EBA) published banks’ top ten credit exposures by region. This dataset enables us to assess whether our findings, based on the correlation of equity prices and determined by markets, have any fundamental justification.

To decipher this, we examine the variation in banks’ credit portfolios to different regions. A bank whose credit exposure is equally spread across all six regions would have the lowest possible variation, and a standard deviation of 0. Although this implies that the bank is well diversified, it also means that it is more directly exposed to a shock in any of the six regions. (The six regions we consider are the same as those displayed within our Minimum Spanning Tree and include the US, UK, Japan, China, euro area and Switzerland). We would expect this to be reflected by a higher average correlation in terms of equity prices, putting it closer to the centre of our MST.

BNP Paribas has the most diversified credit exposure out of the banks for which the European Banking Authority provided data. This supports our findings based on the correlation of equity prices, with BNP Paribas being the most highly correlated so far this year. At the heart of the global banking system, the French lender may be the difference between a regional and a global banking crisis.


Refresh the chart in your browser | Edit chart in Datastream