Connect with us

Capital Markets

It’s back to the classroom for DTCC’s robots

And other challenges of implementing new tech, from blockchain to machine learning, told by I.T. architect Rob Palatnick.



Innovation in digital finance is hard to implement at scale. Just ask the U.S. Depository Trust and Clearing Corporation. Its robots are being required by the human-resources department to take classes on ethics.

This isn’t a question of programming: H.R. wanted the robots to actually take the class alongside humans.

Robert Palatnick, New York-based managing director and DTCC’s chief technology architect, says the robots – which in this case are software, not physical machines – were meant to emulate people’s manual tasks. For them to do so, the robots needed identities within DTCC’s H.R. system, just like employees. But that system then required all identified workers to be trained in compliance and ethics: the H.R. department’s systems didn’t have a means of separating human employees from the machines.

“It’s a cultural issue we’re unpacking,” Palatnick said. “We’ve gone through months of detangling H.R. from making every robot go to class.”

The reality behind innovation
It’s an amusing anecdote (well, maybe not to DTCC’s engineers) but one that highlights the challenges of adopting cutting-edge technology. In this regard, DTCC is an industry bellwether: it is the global financial system’s preeminent post-trade utility, having processed $1.6 quadrillion worth of securities in 2017.

It’s a problem he says afflicts many financial institutions. “Other firms are not seeing the full efficiency gains because they spend a lot of time debugging their robots,” Palatnick said. “It’s a whole new thing that needs human support.”

That doesn’t mean innovation isn’t progressive or helpful, but new solutions need to be carefully integrated into environments that are truly prepared to receive them.

That is often slowed down by the realities of existing business models. DTCC has been here before. For example, it led dematerialization of securities markets starting in the 1980s. Getting rid of paper transformed the entire market, leading to the first phase of automation on Wall Street. But it took years.

“The tech is at different levels of maturity,” he said, noting that cloud computing is now about 15 years old, while distributed ledger is only 10 years old. These technologies lack standards or compatibility. And sometimes they just don’t fit business as usual.

For example, security measures for data centers and critical systems typically involve shutting them down for testing – but this doesn’t work on multi-client cloud services. When every aspect of a function is automated, as it is in the cloud, then making changes involves more than just rewriting the application software: it requires wholesale changes to the entire infrastructure.

Blockchain barriers and bonuses
Perhaps the biggest challenges involve DTCC’s blockchain projects. The firm embarked on two proofs of concept. The one that worked was for a mainframe-based trade warehouse of credit-default swaps: declining volumes made the cost of storing the information too high. DTCC put the data on a distributed ledger and turned off the mainframe: success.

But in the case of repurchase agreements, DTCC was unable to retire the old platform. What looked interesting during the PoC stage turned out to be impractical in real life, because a new DLT platform containing same-day repo transactions would still have to be linked and reconciled with client systems that managed the underlying Treasuries or other debt instruments.

The use case, instead of allowing DTCC to turn off legacy processing, added a new layer. And the business value of adding such a layer didn’t make sense unless DTCC replaced its entire fixed-income core system.

“This went from a small project into requiring us to replace the entire ecosystem,” Palatnick said. “Finding the right business case that the industry can agree on is the best opportunity, and it’s where DTCC can help take the lead.”

Although the repo project was canned, DTCC has gone ahead with advancing its DLT capabilities. This summer, with the support of Accenture, it wanted to test whether blockchain could handle the scale of processing equities transactions that DTCC does in a typical day.

Palatnick said his team wasn’t sure if the DLT vendors would pass the test. Bitcoin processes seven transactions per second; Ethereum does up to 15 TPS. DTCC’s existing systems process 6,300 equities transactions per second.

But DLT vendors Corda and Digital Asset both hit the target.

“This doesn’t change the question of business cases, or whether the industry is ready to move to distributed ledgers,” Palatnick said. “But it does show blockchain can actually scale.”

Despite the many setbacks with innovative tech, Palatnick says firms such as DTCC continue to experiment with them because the end game is powerful: “If other firms stop building their own reconciliation models and tech stacks, and instead opt to be just a node on a standardized network, then the economic magic starts to happen.”

And while he doesn’t see blockchain replacing DTCC’s existing equities infrastructure any time soon, he says the firm is exploring how it could be used for corporate actions, registration and issuance.

DigFin direct!

Register to receive DigFin's newsletter

  • Hauptseite
  • Grocery Gourmet Food
  • It’s back to the classroom for DTCC’s robots