Reviewing Industry best practice for reverse stress testing

Reviewing Industry best practice for reverse stress testing

By Assad Bouayoun, Senior XVA Quantitative Consultant at HSBC.

Assad, can you please tell the Risk Insights readers a little bit about yourself, your experiences and what your current professional focus is?

I am a senior quantitative finance specialist with a focus on total valuation including XVA, risk, stress and reverse stress testing with more than 15 years of experience. I designed industry standard hedging and pricing systems in equity derivatives at Commerzbank, in credit derivatives at Credit Agricole, and in XVA at Lloyds, RBS and Scotiabank. I am now participating to the development of a new library at HSBC.

During these different projects, I integrated new technologies (Cloud, GPU, and QPU), new design (parallelisation using graphs), and new numerical methods (AAD). I am also leading an effort to leverage quantum annealing for XVA reverse stress testing.

What, for you, are the benefits of attending a conference like the ‘Stress Testing Europe Summit’ and what can attendees expect to learn from your session?

Stress testing is not only a regulatory requirement but also a different way of observing the risks taken by a financial institution.
To benefit fully from the stress testing exercise, you must be able to interpret correctly the results and put in place the right mitigating mechanisms. By coming to this summit, you will be able to learn from industrial experts and share experiences around these topics and ultimately make your institution safer. For example a business line putting in place mitigating mechanisms to reduce its contribution to the stress testing results, should be able to benefit from acting on it.

Can you provide some insight into the best practice and benefits of reverse stress testing?

The idea is to lay the foundation of a more quantitative and systematic approach to reverse stress testing. The bottom up qualitative analysis performed to reveal hidden vulnerabilities is necessary but not sufficient.

Regulators are specifying stress scenarios with simplifications that could create a false sense of safety. A drop in house price may not affect in the same way two different banks, as one may over hedge this exposure and not the other. But to avoid a combinatorial explosion, a number of arbitrary choices are usually made in relation to the level of each shock, their combination and the time horizon. These assumptions although necessary, limit the effectiveness of this technique. It is, for example, difficult to determine the correct combination of realistic stress scenarios causing maximum loss. A correlation analysis can help but will not be sufficient for the total value of a derivative portfolio which is a highly non linear function of those scenarios.

Finding the combination of scenarios maximising the loss is a difficult optimisation problem (non deterministic polynomial time). I show in my presentation how quantum annealing (using quantum computer) can help solve this problem and deduce the worst loss and the corresponding set of scenarios.

What are the main benefits of quantitative reverse stress testing?

Reverse stress testing helps your institution prepare for the bad time.

By actively managing and planning ahead, one can reduce the level of loss in case of materialisation of the worst case scenario.
There will be always a worst case scenario, but its severity can be reduced considerably by finding the right mitigation.

Reverse stress testing is not new in nature. The qualitative aspects consist in identifying the different blocks of risk and their dependences have been already extensively documented. It is a necessary step to build the worst case set of scenarios.
A more quantitative and systematic approach with less assumptions can not only help the regulator preventing another banking crisis, but also help the bank better stress manage themselves.

In credit derivative for example, it is a market practice to build a very pessimistic scenario corresponding to all counterparties defaulting at the same time. It gives usually a maximum loss. By modifying the book such as the profit and loss are offsetting when all counterparties are defaulting, we are able to reduce this worst loss considerably.

The extra cost is smaller and the catastrophic event is less severe. But the main outcome is a reduction of the cost of funding the regulatory capital because the mitigation has decreased its required level.

How do you see the risk landscape evolving over the next 6-12 months?

Financial stress testing and reverse stress testing will become a dominant part of the arsenal built by the regulators to protect the economic stability from the eventual distress of one or a series of financial institutions. It will be a useful complement for senior managers and heads of desk who must assess the size of their provisions and manage their tail risk.

For this financial institutions need to seize the moment and embrace some exciting opportunities, most of them linked to recent innovations in technologies (GPU based cloud computing, quantum computing) and in numerical method (AAD on GPU) increasing the compute power while decreasing its cost.

There are multiple consequences. I see more data obtained at a higher frequency available to decision makers. In turn, this will enable them to solve new types of problems like capital allocation and funding consumption.

The Modern Risk Manager will then have to adapt by being more specialized and also by being more IT literate. It means feeling comfortable in a big and fast data environment and being able to prototype new data analysis functions that can help identify and quantify risks in any part of the business.


We asked Assad some informal questions… 

What would you say is the highlight of your career so far?

It is difficult to say. I enjoyed a lot the creativity just before the credit crisis. Besides the acceleration of large scale xVA project development in major banks and the consequence in term of software design change, renders even more interesting working in this area.

I am currently working on a new xVA library for HSBC and participating to the redefinition of the new software architecture. That is the next highlight for me.

What three items would you take with you if you were stranded on a desert island?

o   My development computer with an internet access probably by satellite and all my data and software;
o   a renewable source of energy that could be consumed by my computer; and
o   a ball (football or rugby, I will be hesitating a lot).

 If you had not taken the career route to become a financial risk professional, what would you be doing right now?

I would be probably working on a simulation engine but for aeronautics. Funnily, it is really similar in term of modelling and technology.

One of my best friends worked in this area and we compared different aspects of our work, in particular best practices, the organisation and the resources.

A striking difference was the how unstable the banking environment can be compared to the aeronautic industry