Digital banking: Clouded thinking

Walk through the main trading floor of any global investment bank and amid the dizzying rows of flashing screens and the intoxicating buzz of wheeling and dealing, there lingers a thick air of fear and paranoia.

Such are the demands of financial regulators the world over that almost everything that goes on in there is recorded and monitored by compliance teams trying to safeguard the bank from illegal securities sales and trading activity.

From phone conversations, to emails, to instant messaging and social media use, most of what is said and done is more closely watched today than ever before, oversight that is sucking much of the once ubiquitous buzz from these floors.

The financial crisis and the damaging scandals that have broken since in the interbank and foreign exchange markets have made this Orwellian-like surveillance a new imperative for banks, where some of the most sophisticated technology is being employed to ensure their staff stay the right side of the law.

But what if this technology and the rich flow of data that comes from thousands and thousands of phone calls, emails, messages and social media use could be used for a different, potentially more lucrative, purpose?

Indeed, what if investment banks could use the masses of data they accumulate to help ensure they win more deals, to enhance their view of and raise client profitability, to map out specific market and company exposures, and ultimately deliver to clients precisely what they want, and when they want it?

Such blue-sky thinking may sound fanciful but, faced with the twin challenges of driving profits and reducing costs, it is exactly the kind of new thinking that any global investment banking chief needs to be engaged in.

This is the realm that technologists call ‘big data’, where bulbous-brained data scientists and architects use advanced analytics and technology to extract as much insight as possible from all the tera, peta and exabytes of data that is generated daily by the increasingly digital world we all live, work and trade in.

Some of this information within investment banks is structured data such as capital markets transactions, trades, financial markets data, emails, documents, client queries and more. But much of it is unstructured data that is gleaned from other sources that have exploded in use in recent years such as blogs, Facebook posts, tweets, smartphone apps, electronic sensors and pictures, and YouTube video clips.

But the real value is not in the wealth of data itself, but in what investment banks might be able to do with it.

For years the technology titans of Amazon, Google, Facebook, eBay and even Twitter have blazed a trail in using such analytical techniques to track consumer behaviour and propel their businesses forward at a terrific pace. And yet it is only recently, if at all, that the top tier of investment banks, and the executives who run them, have begun to embrace data analytics in a similar way.

“Automation is inevitable,” says Paul Walker, co-head of Goldman Sachs’ technology division. “Lest you doubt it, how many people are actually selling US equities over the phone anymore versus how many that were in say, 1990?”

He adds: “The coupling of automation for fulfilment with data is incredibly important to the success of our industry. We think that an investment bank that embraces that successfully will inevitably be the winner in the 2020s.”

Digital dilemma

No one global investment bank could claim to have successfully cracked the digital challenge thus far, but look a little deeper into the global markets businesses of some firms and there are examples around how they are thinking about and using big data and analytics to try to do what they do better, faster and far more efficiently.

Some investment banks, says Rupesh Khendry, head of worldwide capital markets at Microsoft, are already using big data analytics to almost immediately assess the impact of the escalation in geopolitical risk on portfolios and their exposures to specific markets or asset classes by combining structured data with the throng of unstructured data that comes from tweets and news.

Similarly, Walker says that at the centre of Goldman’s risk management processes and technology is SecDB, an enterprise-wide database and pricing system, which can model the effects of dynamic market conditions and, in real-time, calculate financial exposure relative to investment positions. 

“This modelling tool considers all of financial positions across the firm and allows for us to run scenarios to improve our understanding and better manage our risk profile for the firm and our clients,” he says.

The use of big data analytics can also be seen in other front and back office areas such as in credit line approval in securities trading; real-time economic or inflation indicator analysis; collateral management; credit valuation adjustment calculations; and advanced mortgage analytics.

But much more can be done. Most investment banks still rely on overnight batch data to make trading decisions, which means their risk models constantly rely on out-of-date data. By using big data analytics in real-time, they can make better trading and risk calls, safeguarding themselves from collapse.

Many investment bankers are wary of this, however. They know the wealth of information they have at their disposal; they also know that the laws surrounding how they can use that data are undefined.

This remains a banking industry under immense regulatory pressure. Investment bank chiefs will err on the conservative side of the legal divide whenever there is a potential issue.

“The difficulty of meshing the predictive ability of your data for customers while maintaining legal and regulatory obligations is one of the main difficulties in bringing these client-facing big data techniques into institutional finance. Implementing appropriate controls in finance around data is one of our most pressing and interesting problems as an industry,” says Walker, although this is an area that Goldman Sachs is looking into with interest.

But there’s no reason, for instance, why FX swaps sales desks shouldn’t be alerted to a corporate client which might be increasing the volume of invoices it issues in a certain currency, and therefore may need to hedge that exposure.

“There is a lot that we can do to enhance the way we do business as investment banks by using all the data that is available at our fingertips,” says Jason Batt, co-head of Deutsche Bank’s markets and electronic trading group in London.

“In electronic trading there is a huge amount of focus on data because it is well understood that this is what is going to take our business to the next level. We’re starting to see the top tier of investment banks increasing their client penetration, and so each investment bank needs to look hard at how they are optimizing their data usage because if they’re not doing that, others will.”

Batt says Deutsche is using data analytics and automation to try to provide a better and faster service in real-time to its institutional clients.

The bank is developing an analytical system that ultimately enables them to generate better intelligence on historical individual request for quotes from clients, and engage in correlation analysis on a client-by-client basis on the trades that it did and did not win. A single basis point can often define success or failure, and with this system Deutsche will be enabled to play around with the pricing infrastructure, creating pricing tailored to a client, and offer a trade to that client instantaneously where before it would have taken longer.

Underpinning this is the quality of data the bank has access to, and even though optimizing the use and management of data is seen as critical, most investment banks are actually still failing to manage it well and this can be expensive.

Mark James, a partner at consultant Oliver Wyman, says that data capture and aggregation are frequently not fit for purpose. Worryingly, banks’ management still make decisions based on data that is often incomplete, inaccurate or dated.

“The results are predictable…mispriced products, limit misuse and the incorrect allocation of capital all result from poor data,” he says, adding that under their analysis bank profits are hit by up to 10% as a result of poor data management.

That’s a staggering hit to take, but what’s just as surprising is how poor some investment banks are at understanding the profitability of their clients.

“Despite major investments in information technology for their capital markets units, few banks have a clear view of revenues and expenses at the client level,” says Kevin Buehler, director at McKinsey in New York.

McKinsey says at least seven of the top 10 investment banks in capital markets have recently undertaken initiatives to understand client revenues and expenses better to try to drive cost savings and balance sheet reduction.

And big data analytics can no doubt play an important role there.

“While profits from proprietary trading will effectively disappear, banks have the opportunity to defend their bottom lines through the informed management of client relationships,” says Buehler. “That transformation starts with data and technology, creating a unique identifier to track each client’s revenues and expenses.”

The net result of doing this, says McKinsey, is that banks could potentially expand the bottom line of their capital markets businesses by 10%to 30%.

Embracing change

For investment banks to truly crack the use of big data requires engendering a different culture, one that is willing to embrace change and innovation, is certainly one of the main challenges.

“Some people are very threatened by change, even if that change is ultimately for the greater good,” says Batt. “The challenge we are all facing is whether we can change that behaviour quick enough to get the efficiency we need and the profitability given the changes in market structure, compression of spreads, and the increased cost of business under the regulatory overheads we all have.” 

The cultural challenge investment banks face here is acute and deep-rooted, something that Sean Park, founder of Anthemis and a former investment banker at Dresdner Kleinwort Wasserstein, encountered a decade ago.

In 2005, Park and colleagues launched what was then the first digital markets division in the investment banking industry, only for it to fall apart, along with the bank, not very long afterwards. Much of what they tried to build then was exactly what investment banks are trying to do today in data analytics and technology.

However, Parks says the idea was “culturally puked up”.

“To be fair to the senior people there, there was little push back to the long-term thesis and trends that were happening,” says Park with the benefit of hindsight. “But investment bankers and traders can be quite cynical and short-termist. Their view was that this is the next guy’s problem and let’s get going while the going is good. We know where that led.”

He adds that this cynicism all too often still exists but that increasingly leadership teams of these institutions get what is going on, “they are smart people, but more often than not it’s a case of them saying: ‘I get it, but what the fuck do I do?’”

Park says there is also a paradox of power in large financial institutions. “The CEO may say ‘let’s do this’ but the ability to actually get it done, and done well, resides a few levels below. And down there in the ranks, there’s a lot of fear because there’s a pretty strong realisation that the world is changing fast and that’s provoking trepidation and the simple question: when is it coming to me?”

Changing internal cultures will take time but investment banks can be ruthless when change is impeded. But another question is whether internal culture can change enough to attract some of brightest brains in the data analytics field.

While there has been a run of large international banks appointing chief data officers and architects in the past 12 months or so, attracting such talent en masse is proving tricky. The lack of an open and highly innovative culture that a Google, Amazon or Facebook have is clearly one reason for this, another is that investment banks are, for once, apparently being priced out of the job market.

“If you look at the salaries being offered by companies such as Amazon, Google and Facebook, they are far outstripping anything the investment banks are offering for real leading talent. The amount of equity being given out to attract the best is considerable,” says Vincent Kilcoyne, UK and Ireland capital markets industry lead at SAS, one of the largest business analytics software providers.

That may be so given the remuneration pressure investment banks are under but some have plenty of technology specialists. Goldman, for example, has a technology division of 8,000 people, more than a quarter of all its employees, including its ‘Strat Business Unit’.

Strats, as people working for the unit are known internally, work right across the firm’s trading and sales, banking and investment management divisions. In trading and sales specifically, some of them are the type of rocket scientists that sit on trading desks, creating cutting-edge derivative pricing models and developing empirical models to provide insight into market behaviour. Others are developing automated trading algorithms.

Their backgrounds include science, mathematics, engineering as well as economics and finance.

What’s interesting is that these strats are ascending the ranks, pointing to the strategic direction Goldman is taking.

One of the most recent and high profile examples of this was the promotion of Martin Chavez – who joined Goldman in 1993 as a senior energy strat – from co-chief operating officer of Goldman’s equities business to chief information officer late last year. Chavez is known for his technological and analytical skills, having overseen the creation in the 1990s of an internal software platform ‘Marty’, since renamed Marquee.

In an internal memo announcing the appointment, Lloyd Blankfein, chief executive of Goldman said: “As markets and regulations evolve, Marty will lead our efforts to ensure that we deploy effective and innovative technology to address our clients’ needs.”

Fundamental challenge

If there is one fundamental challenge that the top tier global investment banks face in being able to do what they want and need to do in big data, it’s the fact that none of them have built their systems with this type of advance in mind.

“If you take a look at Twitter, which is making a killing in the big data space, it makes sense that they do because they have written everything in the last three years and they have the systems for that,” says a senior investment banker.

“I wish we had the application but we don’t. We have thousands of applications, systems and databases and bits of code. And so figuring out where you apply the observation such that you can then apply the big data analysis requires a very flexible approach and that is one big barrier the industry faces.”

The complex and often outdated data architecture in most, if not all, investment banks is a particularly acute problem, frustrating their ability to analyse and extract deeper insight and value from the data they have, in the way they want.

“The data problem is actually compounded in many of the investment banks because it’s multiplicative, which means they face some pretty serious problems around even tackling the most basic of questions, or doing the most basic analysis. Around 99% of the work that they are doing is XL based too,” says one technology adviser and big data specialist.

He adds: “If they want to get close to big data, where it’s almost NASA-like in its sophistication, investment banks will face multiple challenges first. Achieving that will involve centralization of data, of analytics and of reporting. And as I see it many investment banks are very far away from that.”

Basil Qunibi, CEO of Novus, a portfolio analytics and intelligence platform for institutional investors, and a former investment banker at Merrill Lynch, gives a simple but striking example of this: “If any investment bank says they are investing in big data, ask them to pull up their aggregate exposure to Apple and ask them to show you how it breaks down by product.”

He adds: “So if a bank has 1000 products and 300 are invested in Apple, they should, ideally, have a centralized database that shows the number and value of shares across all of these products, enabling the bank to aggregate its total exposure to Apple. However, I’d bet that most investment banks would not be able to answer that question using the simple analysis I have just given. And we believe that’s pretty shocking!”

In the last couple of years there have been a couple of big changes in the way the mathematics and technology works that help support investment banks in being able to search and analyse large data assets and more. The first is the emergence of scalable open-source infrastructure, such as Hadoop and Elastic Search. The second is machine learning, a form of artificial intelligence.

Hadoop, which is being used by Morgan Stanley, Bank of America Merrill Lynch and Goldman, among others, is an open source software that enables the processing of large data sets across clusters of commodity servers. It is designed to scale up from a single server to thousands of machines. Elastic Search is an open source search and analytics engine that too enables the exploration of data.

Using both “allows us to make these very fast, unstructured, semantic queries across the data. That’s very important,” says Walker.

The use of AI or machine learning in investment banking could be a game-changer. AI already powers Google’s algorithms, friend recommendations over Facebook, and can be found in aircraft landing systems and self-drive cars. And some hedge funds, such as New York-based Rebellion Research, are using AI to manage investment portfolios. They launched their first strategy back in 2007.

Walker says that Goldman have had some great luck with applying machine learning to its network support centre, which can materially impact returns.

“Machine Learning can improve our return on equity by increasing revenue or decreasing expense,” says Walker. “For example, by taking the exception rate for network alerts down by more than 50% with machine learning, we can run a more efficient plant.”

Vital investment

There is a growing belief among investment banking executives that big data has the potential to transform the industry in ways they may not have thought were possible. Making the required investments now in the systems and technology that is needed is vital.

“The large capital markets firms increasingly do recognise that the data assets they have offer a way to help tackle a lot of the cost challenges, the regulatory and control challenges, and the revenue and growth challenges they have,” says Jared Moon, principal in McKinsey’s London office.

Indeed, for Deepak Goyal, partner and managing director of Boston Consulting Group, investment banks that take a data-led advisory approach to client business are on the right path to reviving profitable and sustainable growth.

“Our work with wholesale banks has demonstrated clearly that concentrating on increasing the share of wallet with existing clients – as opposed to relying primarily on developing new business – and taking a data-led, advisory approach can lead to revenue growth of up to 15% over a 12-24 month period,” he says.

There are broadly three layers to this data investment, says Moon: the foundational layer; the analytical layer; and the revenue enhancement layer.

This third layer is essentially about managing client profitability and using the information on what an individual client does with the investment bank, and particularly on their single dealer platform, to predict the next product they should buy, or offering services such as bespoke valuations.

Moon says McKinsey has done some rough estimates on what the payback looks like for making such investments, and on the foundational layer alone he says that over three to five years, for a typical top-tier capital markets firm, it could generate upwards of a billion dollars, across revenues, risk-weighted asset reduction, and productivity front to back.

That’s got to be an attractive revenue uplift for any investment bank. But it still may not be enough to make big data a top priority investment for a bank CEO.

It should though. “When looking at the future of your business and what will drive it forward, I think of growth, regulation, the cost base, and capital, liquidity and risk management are some of the top issues,” says Walker.

“I don’t think that big data gets added to the list. I see big data is a technique that allows you, as a bank CEO, to address those four issues more effectively.” 

And those investment banks which embrace big data are better set up for future success much more than those that do not, says Microsoft’s Khendry.

“Whether it’s the ability to cross-sell services, improve customer service or come up with smarter, data-driven decisions around trading and risk/exposure management, the industry recognizes that data is the new currency,” says Khendry. “It’s something that the smart banks are taking advantage of.”

However, as Ron Papanek, head of the risk metrics alternative investments business at MSCI, a provider of indices, portfolio risk and performance analytics to global financial institutions, points out, investment banks need to start looking outside for innovative solutions to stay that step ahead.

“It’s very difficult to stay cutting edge, let alone bleeding edge, when you are building everything internally yourself,” he says.

Adrian Crockett, CEO and co-founder of technology firm Pellucid Analytics, and a former managing director and head of strategic finance group at Credit Suisse, shares this view.

“The problem is that a lot of the technology here cannot be created inside a bank because there is a cultural misalignment they can’t get over. IT is directed to focus on regulatory compliance, service level agreements, and cost. They’re not measured on how much value they extract from data.” 

He adds: “One CEO said to me, ‘I understand the cultural issue, but what do I have to do to eradicate that?’ I said, ‘First, you have to differentiate between what’s utility and what’s value-add. You’re spending far too much on utility investments and that’s why your cost base is so high. Look for ways to externally source the utility functionality, and then overlay your proprietary elements, rather than building from the ground-up.’” 

A simple example of where they are spending too much money in investment banking is on the pitch-book – the bane of any junior investment banker forced to spend hours compiling them and any senior banker waiting for it to be done. 

Crockett and his team at Pellucid worked out that the average pitch-book costs $40,000-$70,000 to create, given the expensive personnel hours and other factors. They developed and launched a sophisticated but simple digital pitch-book platform that slashes the time and cost required.

That kind of technological advance is something that at least investment bankers should be happy with, and not fear.