Fintech is triggering a profound rethink for financial institutions, from retail to investment banking. However must this imply the banking model is broken beyond repair thus leading to massive headcount reductions? At least for investment banks the answer is not so clear cut. The combination of services and infrastructure traditionally housed under one roof – underwriting, research, sales & trading, supported by large back office operations, and monitored by compliance systems – will remain at the sector’s core. This combinatorial model works because it’s diversified, can best withstand bear markets, benefits from technological synergies, and it’s the mix of products and services clients value. However each component will change dramatically. Whilst not leading to massive headcount reduction, it will require a change in skill sets and attitudes.
To dig deeper, let’s first review the influence of technology on the core components.
Jeff McMillan, chief analytics and data officer for Morgan Stanley’s wealth-management division, “… when I talk to financial advisers, they’re always like, ‘Is this going to put me out of business?’ … that’s always the big elephant in the room. I can tell you factually that we are a long ways away from that.”
Research and Trading:
The speed and accuracy of research will be vastly improved with the application of big data and analytics. While detecting tiny shifts in market data streams is known to power high-frequency trading, a bit closer to ground level traditional research is increasingly being impacted by AI too, in particular, “behavioral intent.” Here digital intercepts of consumer activities are aggregated into large data sets, analysed, and assessed versus market expectations. An early example occurred in 2010 when UBS Analyst Neil Currie accessed satellite imagery to monitor activity in Walmart parking lots, running the data thru a mathematical regression to translate it into customer activity for better earnings forecasts. Seven years on and with90% of the world’s data produced in the last two years, the applications have become far more ubiquitous. Every day Sqreem Technologies (a Singapore company I advise), tracks the digital footprint of 3 billion consumers as they engage with over 150,000 brands, capturing 75% of online behaviors across 40 countries. Coupled with powerful pattern-detection capabilities, the information can be transformed into asymmetric trading signals for both fundamental and program-driven strategies.
Think of big data-driven research as the backend of analytical information processing, and Robo Advisors as the front end delivery mechanism to salespeople and their clients. PC and mobile interfaces dynamically display portfolio valuations and exposures, along with system-generated investment recommendations tailored to a specific client’s financial goals and risk appetite. Morgan Stanley predicts that Robo Advisors will manage $6.5 trillion by 2025, about 5% of the world’s wealth.
Blockchain is a distributed ledger technology (DLT) wherein all network participants can have separate copies of the entire record of transactions on the network. Legacy architecture in financial services, by contrast, is comprised of unique centralized databases, requiring the need and added cost of intermediaries to verify transactions between them (e.g. the role of SWIFT in international wire transfers). DLT removes the middlemen by enlisting the network of users to verify transactions. However, updating a DLT network takes time; currently Bitcoin’s DLT processing speed is 7 transactions per second, compared to Visa’s centralized database which routinely handles more than 2,000. This limits positive ROI DLT installations to lower volume/intermediated environments.
Alas multiple efforts are underway to upgrade DLT for large scale adoption. One example, Singapore’s Zilliqa, is developing a high through-put DLT platform to handle the thousands per second capacity required by a Visa but with much lower merchant fees, and with plans to scale further. To be in a position to benefit when the higher-volume platforms become fully operational, key players are launching lower-volume pilot programs. In May Citibank and Nasdaq announced an arrangement to allow investors holding private company securities on Nasdaq to be able to use Citi’s cross-border payments facility and blockchain to buy, sell and settle transactions, bypassing the heavy paperwork typically associated with this type of transfer. According to Adena Friedman, CEO at Nasdaq, “through this effective integration of blockchain technology and global financial systems, we can realize greater operational transparency and ease of reconciliation, which can have profound implications for outdated administrative functions in the capital markets.” In Singapore, a research piece from Deloitte and the MAS (the country’s central bank and financial regulator) details how under “Project Ubin”, a joint initiative which includes several banks, a payment system prototype is being developed using DLT in which banks can exchange currency with one another without lengthy processing times, expensive processing fees, or intermediaries.
“Thus far investment banks have been reluctant to participate in cryptocurrencies, but expect that to change because where the buy-side leads, the sell-side will follow.”
Compliance is being disrupted as algorithms and other software driven programs can be written to detect and report anomalies much more efficiently than human compliance officers. As always the key challenge here is to reduce the number of “false negatives” (an undetected breach) to as close to zero as possible while limiting the number of “false positives” (a non-breach identified as an investigative-event). While false negatives can be catastrophic like when rogue trader Nick Leeson caused the collapse of Barings Bank, false positives have a cost too in terms of staff reporting time and investigation. Here’s where machine-learning comes into play to self-correct over time, dynamically adjusting detection algorithms based on experience, aiming to keep everything within a controllable cost/risk limit. This is a major issue and the role of data analytics will continue to expand as the world’s top investment banks were fined a total $43 billion over the last seven years for failures in customer reporting.
Bitcoin has gone somewhat mainstream since its introduction eight years ago, however less well known are the subsequent cryptocurrency introductions. In order to fund a project or business, new cryptocurrencies are created by an offerer, swapped with investors for cash or other cryptocurrencies, and listed on exchanges to provide liquidity and price transparency. This type of transaction is known as an Initial Coin Offering (ICO), and its use is growing rapidly. Already thru mid-October of this year $3.3 billion has been raised from 203 deals, or 33x the amount raised all of last year.
Although the landscape is fraught with some growing pains, ICOs are no longer a fringe niche, with existing funds like Metastable and Polychain Capital setup to invest in them. Another fund preparing to launch, Global Public Offering Fund, boldly proclaimed that “the U.S. IPO market is broken”, positioning their offering as an alternative to capitalize on the dwindling supply of smaller IPOs, pointing out that only 18 companies completed IPOs that raised less than $50M last year, versus 557 in 1996.
Thus far investment banks have been reluctant to participate in cryptocurrencies, but expect that to change because where the buy-side leads, the sell-side will follow.
“Capital markets professionals at some point will feel the pinch of ICOs and other crowd funding techniques as they snatch profitable underwriting opportunities from their pipelines. Perhaps this more than anything else is what Jamie Dimon is reacting to when he predicts the end of Bitcoin (really more like cheerleads for it). Rather than sound the alarm for regulators to save them again, bankers will be better served by focusing on retraining here too.”
Headline numbers post-credit crisis have mostly reported reduced banking employment. But how much of the story can be traced to overall business conditions? More importantly what will be the impact of the inevitable march of technology?
According to research from JP Morgan, revenues from investment banking peaked in 2009 at $207.7 billion. After the credit crisis they declined sharply and seem to have stabilized around $140 billion. Not surprisingly industry headcount also gapped down, and as the chart below reveals has subsequently become relatively stable too.
Today employment in the sector is comparable to levels from 2005-2006 when revenues were also similar. So although technology has certainly advanced rapidly since then, the effect is not showing up on revenue per employee. Revisiting our components let’s see why.
In client coverage, that same Morgan Stanley who forecast robot’s managing 5% of the world’s wealth by 2025, in May announced their 16,000 brokers were getting a machine-learning algorithmic boost. Quoting Jeff McMillan, chief analytics and data officer for the bank’s wealth-management division, “technology can help them understand what’s happening in their book of business and what’s happening with their clients, whether it be considering a mortgage, to dealing with the death of a parent, to buying IBM … we take all of that and score them on the benefit that will accrue to the client and the likelihood they will transact.” On the issue of headcount he went on “when I talk to financial advisers, they’re always like, ‘Is this going to put me out of business?’ … that’s always the big elephant in the room. I can tell you factually that we are a long ways away from that.”
In the area of transaction settlement, blockchain will cut out the intermediaries, but note those middlemen being cut out are outside the bank. It’s the SWIFT’s of the world that need to totally reinvent themselves. And they’re trying! “22 additional global banks join the SWIFT gpi blockchain proof of concept.” Internally what banks need to do is train their database personnel in DLT … maybe some will be replaced, but replaced by individuals with the desired skills, not by the software itself.
Compliance is a special case with a unique set of factors. On one hand computers by necessity are the core of the system because humans can’t even dream of performing the required calculation volumes. On the other hand, the systems they’re monitoring continue to generate an increasing amount of information. Add to that increasingly strict regulatory regimes and according to Scott Towend, team leader for compliance and risk at recruitment firm Randstad, “demand has been, and will continue to be, strong for qualified and experienced risk and compliance candidates.”
Capital markets professionals at some point will feel the pinch of ICOs and other crowd funding techniques as they snatch profitable underwriting opportunities from their pipelines. Perhaps this more than anything else is what Jamie Dimon is reacting to when he predicts the end of Bitcoin (really more like cheerleads for it). Rather than sound the alarm for regulators to save them again, bankers will be better served by focusing on retraining here too. Although underwriting an IPO shares some similarities with ICOs, it’s only at the highest level and the details of executing — both technically and pragmatically — are quite different (for some insight think of old-line white shoe investment bankers versus millennials, and how and where they communicate differently).
For AI’s impact on research and trading professionals, I’ll quote from Nick Granger, CIO of Man AHL, as it also provides an appropriate overall concluding remark. In 2014 Granger, a Ph.D. in mathematical logic and senior portfolio manager at the time, started an AI-program trading small amounts. Due to its success, today that program (AHL Dimension) manages $5.1 billion, and four Man Group funds collectively manage $12.3 billion incorporating AI. In a Bloomberg News piece from September, when asked about the prospects for AI completely automating their business any time soon, Granger responded: “The idea that the humans will just disappear and would be banned from the process is just not right … It’s just that they move to different tasks, to higher value-added tasks. We need smarter humans than we did.”