Bank risk manager of the year: Natixis

Risk Awards 2023: Revised risk management framework and front-office collaboration help bank withstand 2022 volatility surge

L to R: Alexander Cherdel, Michael Haize, Joel Benaroch, Pascal Amiel and Arie Boleslawski

Image: Fabrice Vallon

History is our greatest teacher. But in risk management, one potential pitfall of this is always to be preparing for the last crisis.

Instead, Natixis Corporate and Investment Banking (CIB) has used the lessons of previous risk management events to build itself a new, dynamic framework – one that has already proven its merits in the wake of the 2022 invasion of Ukraine and the spike in both inflation and interest rates.

One trigger for the overhaul was the bank’s €260 million ($280 million) loss, sustained in 2018, when it expanded aggressively into complex Asian autocallable products, then struggled to hedge its exposures when volatility and correlations jumped.

In 2020, the Covid pandemic helped accelerate the new framework’s development, powered also by a change of management and a strategic review. The overhaul saw forces from the bank’s first two lines of defence combine.

“This is something we designed collectively between risk management and front office,” says Michael Haize, global head of global markets at Natixis CIB, describing both the aegis for the bank’s new risk management framework and one of the keys to its success.

It passed an early test of its mettle in 2022’s equity market volatility, when the Euro Stoxx plunged to 3,279 at the end of September, almost 25% off its 4,332 annual start, before recovering to a loss of around half that magnitude by year-end.

The whole framework works in an environment where the equity market is pretty bearish, and this is what we want – to generate positive P&L, whatever the market conditions

Michael Haize, Natixis

Combined with the interest rate volatility that typified 2022, this scenario would likely give rise to strong conditions for any bank with a large, fixed income flow business, but not normally for a bank like Natixis that leans toward equity structuring.

In practice, however, results have been encouraging. Revenues from the equities business hit €313 million ($337 million) in the first half of 2022, exceeding an annual run rate of €300 million ($320 million) that was set as a target in the 2020 strategic review.

“The whole framework works in an environment where the equity market is pretty bearish, and this is what we want – to generate positive P&L, whatever the market conditions,” says Haize.

The framework deploys a hedging complexity score to monitor exotic risks and two-to-three-week shock scenarios, combined with a new local stochastic volatility model, using adjoint algorithmic differentiation to efficiently calculate sensitivities at scale.

But at the heart of the new approach has been the close partnership between front office and risk management to develop tools that work for both parties.

“This improved risk framework enabled us not only to better manage our risks, but to do more business,” says Haize. “We want to grow, but in a sustainable way.”

“If both parts are not aligned on how we want to run the business and manage risk, it is not possible,” adds Joel Benaroch, global head of market risk at Natixis. “For us this is the best way to act.”

At Natixis, as at all major European Union investment banks, the large moves in some markets, such as commodities and rates, have triggered value-at-risk regulatory backtesting breaches over the past 12 months. But Haize says the new tools have enabled management to have “a better view of VAR consumption and the global behaviour of our books”.

And of course, the bank is not alone in questioning its capabilities in this respect. “Regulators are asking the same question, and the short answer is – it’s working.”

Hedge complexity score

The new framework’s hedge complexity score came into full utility at just the right time. Designed to monitor risks such as convexity, concentration and correlation – that go beyond the first or second order risk measurements – the tool then distils these into a dashboard that management can readily use.

Where the complexity score for a product or business line is high, the bank can choose how to strengthen the risk management framework, or whether to scale back.

“We started with the equity market in 2020, because with rates that low, the product we could sell was autocalls, some of which can have a lot of complexity,” says Benaroch. “Based on that, we then extended the approach to all the products in global markets activities – fixed income, commodities, credit.”

And not a moment too soon. Haize says there is no doubt that the hedge complexity scoring helped the bank “weather the storm” in 2022.

“This system points out all the products that are in our scope, to make sure all of them could be correctly framed if we had this kind of volatility,” says Benaroch. “We managed to improve some limits and introduce complexity limits where we did not have them before.”

This system points out all the products that are in our scope, to make sure all of them could be correctly framed if we had this kind of volatility

Joel Benaroch, Natixis

Turning complexity into a more accessible scoring system has also improved internal and external communication of risk policy to management, shareholders and clients.

“We are there to provide investment, optimisation and hedging solutions for our clients, and it is important to them to know we have the capacity to keep showing them liquidity, especially in crisis times,” says Haize. “What you don’t want to do with clients is have some kind of stop-go attitude because you are not comfortable with the product – so we have to be very comfortable with the products we keep on our books.”

The changing macroeconomic and market environment has also broadened the range of products that are potentially attractive to investors, adds Haize, moving beyond autocalls and into capital-guarantee or growth and income offerings. What the team calls the “cartography” of risks created by the hedge complexity scoring has enabled the bank to be nimbler with new product offerings.

“This framework creates opportunities for new products to balance existing positions,” says Haize. “It enables us to have a better economic allocation of our capital toward certain types of products. It helps improve the profitability of our business because it helps us know where to allocate capital and liquidity on certain products, improving overall capacity.”

New model roll-out

Another timely development was the full production roll-out of a new local stochastic volatility (LSV) model, which can separately handle the dynamics of implied volatility and of vanilla pricing options.

The original LSV model dates back to the late 2000s and was already in use for pricing on an ad hoc basis since 2012, says Benaroch. But running it across all the existing derivatives books for all the necessary risk measures was another matter.

“The challenge was to stabilise the implied volatility surface, because it is a very complex model, so if the implied volatility surface is not smooth, the model does not run correctly,” says Benaroch. “What is also complicated is to have stable parameters when the market is moving sharply. And, finally, what was challenging was to produce the greeks, the sensitivities at all the other levels, strike against maturity.”

This framework creates opportunities for new products to balance existing positions

Michael Haize, Natixis

To reduce the heavy computational demands, the bank also assembled a team of quants from front office and risk management to develop an adjoint algorithmic differentiation approach to use the LSV model on a large scale.

“We developed innovative calibration techniques and advanced numerical implementation tools that allowed us to compute all those products’ greeks very quickly for each portfolio,” says Benaroch.

This LSV model helps to solve some of the risk management problems Natixis experienced on autocall products when the Korean stock market fell sharply in 2018 – problems that re-emerged across the banking sector at the start of Covid lockdowns in 2020.

Normal modelling assumptions would be based on volatility rising mechanically as prices fall because the same price increment represents a larger percentage of the new, lower stock price. But the size and concentration of autocall books in the options market can cause that inverse correlation to break down because positions were so sensitive to vega – changes in the volatility of the underlying.

“Most autocalls are running put options, so everyone was long vega on the downside,” says Benaroch. “When you start to hit your limits, you have to sell, so it was some kind of one-way market, and not having a proper model for that was not an efficient way to run the books.”

And this time really was different. Even as stock markets fell sharply in 2022 amid the Russian invasion of Ukraine and the surge in inflation, Benaroch says the LSV model remained stable and its outputs were vital for managing the changed market landscape: “It was a pretty strong move to the downside, but we managed to do well, and the model was a key element in that success.”

Natixis has so far this year been able to avoid the pattern often seen in equity derivatives, where sharp falls force banks to pull down the shutters on new offerings while they figure out the risks on their legacy books.

“Because we have a good view of our positions, we can continue doing business with our clients which is what we want,” says Haize. “We have increased our commercial activity this year very efficiently.”

Scenario storyboard

Another new technique that has come into its own in 2022 is a further joint effort – this time between risk management and the bank’s economic research team. Natixis wanted something between the daily stress calculations for market risk, and the one- to three-year horizons used to carry out the supervisory internal capital adequacy assessment process across the whole balance sheet.

The result is a set of two- to three-week shock scenarios, updated every few weeks, which can be applied to trading books as needed, then discussed internally to set risk policy.

“It helps us to have this kind of view so we are clear on what could be the negative outcomes, and we can fine-tune hedging policies – looking at exactly the kind of hedging we have put into place, seeing the kinds of scenarios that could generate some pain on specific products, positions or situations,” says Haize. “It is not a direct action plan, but it helps us to better understand the way our positions shift… It helps us with the tail risk hedge position that we take.”

Those benefits accrue even if the scenario does not match exactly what happens over the weeks that follow.

In practice, says Benaroch, a number of the scenarios the bank developed last year proved highly relevant – including the effects of the conflict in Ukraine on commodity prices, a UK interest rate shock, and a resurgence of Covid in China.

Instead of preparing for the last crisis, Natixis has given itself the best possible chance of predicting the next one.