Bank risk manager of the year: Deutsche Bank

As markets became increasingly jittery in October, Deutsche Bank chose to rein in its exposure - slashing value-at-risk by more than $30 million in three days and demonstrating the benefits of an increasingly liquid portfolio

Stuart Lewis, Deutsche Bank

One of the challenges facing all big banks is how to devolve risk-taking authority to hundreds of trading desks dotted around the world, while still being able to cut exposure quickly. The oil tanker metaphor is often used – it takes a while for these big, complex organisations to change course.

For much of last year, there was little need to put those capabilities to the test. That changed in the middle of October, when mounting losses on a series of bad bets forced hedge funds to mechanically cut exposure, ultimately causing US Treasuries to whipsaw on October 15. The yield on the 10-year bond, which closed at 2.21% the previous day, plunged to an intra-day low of 1.87% at 9.39am New York time, before recovering sharply to settle at 2.14% by the close.

Ahead of that move, increasing market jitters prompted Deutsche Bank to take evasive action. Over a three-day period, it cut its global value-at-risk numbers by almost one third, behaving less like an oil tanker and more like an America's Cup yacht.

"I have never seen us work so convincingly to de-risk our book as we did that week. This was a case of having to get all the business heads on a call; we all knew we were heading into a hairy patch, so it was about getting the best minds in the business together to identify what our collective view was, and how we should reposition to ensure we could pare down our exposure," says Stuart Lewis, chief risk officer at Deutsche Bank in London.

A trader might smirk – the hairy period for which Deutsche was preparing lasted for the three days of its de-risking and volatility subsequently cooled off. In essence, the bank was hedging at the market's low point.

I have never seen us work so convincingly to de-risk our book as we did that week

Quick consensus

From a risk manager's point of view, that is no criticism – the point is that the bank had little conviction about where markets were heading in the coming hours, days and weeks, and was able to quickly establish a consensus, and then to materially change its risk profile. That should be reassuring to senior management, investors and supervisors, and is ultimately what risk management is all about.

It required a lot of collaboration. Heading into the fourth quarter, Deutsche Bank had been loading up on both equities and credit inventory in anticipation of strong client demand over the last months of the year. It meant the bank, which had a short position in US Treasuries in early 2014, now found itself with a portfolio that was long US dollar, long US equities and long credit, while awaiting client interest that had yet to materialise.

Markets had been increasingly jumpy since late September, when Bill Gross left Pimco, oil prices fell and stock markets also started to slide, but positions really started moving against Deutsche Bank around October 10.

Although traders and risk managers alike initially saw this as a temporary adjustment, anxiety was building by the following Monday, October 13. Exposures across asset classes remained within the bank's risk limits but were drifting towards them until, on October 14, Deutsche bumped up against one of its global – rather than asset-class specific – risk limits. At the time, its VAR models were estimating a potential maximum loss of up to $73 million.

Up to this point, the changing risk landscape had been discussed in regularly scheduled weekly meetings of the upper echelon of the corporate banking & securities (CB&S) business, as well as Deutsche's finance and treasury departments and biweekly risk committee meetings, but an ad hoc meeting was called in London on the morning of October 15, involving business heads, traders, research analysts and senior management from the CB&S division.

The agenda was to work out how much risk to take out of the portfolio - and where to cut. The consensus was that increasingly intense volatility could be a precursor to a sustained rout in a variety of assets, and the bank decided to cut quickly and significantly.

"Over the next week-and-a-half, we took the VAR number down by around 40%, with more than half of that reduction coming in the first day and a half," says David Stevens, global head of market risk at Deutsche Bank in London. "That was mostly done by reducing directional equity and forex exposures, as well as cutting back on interest rate positioning, reflecting the rally in US Treasuries. Additional reductions over the following week mainly came from cutting back the credit exposure, which was not as easy to move in size."

It was not only the sheer volume of positions that were taken off that appears remarkable, but also the speed with which the de-risking was achieved, especially in the first 48 hours during October 15 and 16.

"The initial meeting took place on the Wednesday. We told the board what the strategy was going to be and by the Friday we had achieved a very large amount of de-risking. In that time, we managed to take the VAR number from somewhere north of €71 million to €46 million. In my 18 years at the bank I have never seen such a single-minded approach to take risk down so dramatically when called to do so," says Lewis.

Level 3 asset disposal

The ability to liquidate large portions of the portfolio was crucial in Deutsche Bank's response to that acute period of stress, but it was only possible because of the longer-term commitment the bank has had to disposing of illiquid or hard-to-value positions – level 3 assets, in accounting jargon.

"Part of our business strategy in recent years has been to make the trading book far more liquid and I think that you can see that. We have taken our level 3 assets down from north of €90 billion in 2008 to under €30 billion. So October's de-risking is a testament not only to the level of co-operation on that day, but you can look back over several years and see an ongoing process of de-risking and liquidating the portfolio. To what extent this will continue, I would say we are in a new paradigm in which banks have deleveraged quite substantially: we are simply carrying a lot less inventory and we are seeing some large price moves on the back of thinner markets," says Lewis.

Among the less-liquid assets the bank has shed in the past year is Las Vegas hotel The Cosmopolitan, which it unloaded in May 2014 for $1.73 billion. It continues to look for a buyer for its Maher Terminals business, which operates port facilities in the US and Canada.

Elsewhere in the bank's non-core portfolio – representing €45 billion of assets as of the third quarter of last year – Deutsche exited a large piece of its correlation business, after allegedly being frustrated in its attempts to do so in early 2013. The correlation book, which contains legacy securitised credit products originated in the pre-crisis years – such as tranches of synthetic collateralised debt obligations, and nth-to-default credit baskets – was poised for a major transaction two years ago, but German regulator Bundesanstalt für Finanzdienstleistungsaufsicht (Bafin) blocked the trade over capital concerns, hedge funds involved in the business told Risk in November.

Lewis would not discuss the details of that episode, or confirm Bafin's involvement, but he says the bank has traded away a hefty block of this book in the past year.

"In 2014, we successfully reduced our correlation risk by about 30%, so that is quite a large trade. The last number I had seen around correlation risk was approximately €200 million of downside loss on that book, so a relatively small amount in the overall scheme of things," he states.

Performance

Lewis and co are also proud of the bank's performance in the European Central Bank's (ECB) comprehensive assessment and asset quality review (AQR) of the health of 130 eurozone banks, published on October 26. The review found that on a balance sheet totalling €1.58 trillion, Deutsche Bank's modelled core equity Tier I ratio required just a seven-basis point adjustment under the AQR calculation.

"The AQR stress test was a huge data-gathering exercise, involving the collection of reams and reams of data, and we see the results as a resounding success for the firm and demonstrated the great work of the finance and risk divisions. Both our financial and our risk management view of assets was appropriate, especially in the credit space, where we had a €400 billion loan portfolio that only required approximately €100 million of additional loan loss provisions. That was really a small magnitude of error for such a large portfolio," says Lewis.

It's not hard to see why the adjustment was minimal. The bank recognises no single risk measure will offer a complete picture, so tends to use multiple approaches, comparing and cross-checking to look for weaknesses. One example of the benefits is found in a recent review of risk management for the bank's securitisation business.

Traditionally, the bank has hedged its real-estate securitisations directly, using the ABX and CMBX indexes, which reference a basket of 20 subprime residential and 25 commercial mortgage-backed securities, respectively. But, based on a VAR analysis, the business found more reliable correlation when using the AAA- and AA-rated tranches of investment-grade corporate loan securitisations as a hedge, even though the underlying assets are very different.

The higher degree of correlation using these alternative hedges made the switch look sensible, until the market risk team ran an analysis of how those hedges would perform not just under a regular VAR calculation, but also in terms of economic capital and when using stress tests.

"What was important was our understanding that the VAR metric would not capture everything that was going on in the portfolio," says Stevens. "In looking at our other two portfolio metrics, which capture stress-type scenarios to differing degrees, we picked up a growth in basis risk between inventory and hedges in our securitisation and commercial real estate trading portfolios. The VAR – with its 12-month time horizon – did not pick that up, and hence their hedges were working according to the model. However, our other two metrics showed the hedges would not work under crisis-like stress moves. This gave us the ability to discuss whether that strategy was right for the current environment."