The leading knowledge platform for the financial technology industry
The leading knowledge platform for the financial technology industry

A-Team Insight Blogs

JPMorgan’s Chief Data Officer Peter Serenita Gets to the Hub of the Matter

Share article

JPMorgan Chase has taken an “operational hub” approach to centralising reference data for its wholesale business, which has improved efficiencies and reduced costs, with most business units seeing “paypack within one to two years”, said Peter Serenita, chief data officer, JPMorgan Worldwide Securities Services, who shared his experience at FIMA 2006 in London earlier this month. The migration of additional business units to the hubs is continuing gradually, with 14 new business units moving their data operations into the hub infrastructure through 2006.

JPMorgan chose to build its own systems for managing the data five years ago, decommissioning old systems. It created two reference data operational hubs, one in Delaware and one in Mumbai, according to Serenita, who provided food for thought for delegates about the process, from the decision of where to site the hubs and what functions to put in them, through how to handle the migration, to how to measure the results and maximise the cost and operational benefits of such a move.

The choice of Mumbai as the offshore reference data hub for JPMorgan was aligned with a bank level decision to create a global services centre there, and having such an alignment is vital, Serenita said. “If you choose a site that is not a strategic location, you will face problems with the infrastructure – you need desktop support, for example.” Another key decision to make upfront is whether to migrate the functions as they are and re-engineer them later, or re-engineer first. “To lift and drop means having to retrain later, but we did that because everyone currently doing the functions we were migrating was already geographically dispersed and organisationally aligned. We did the migration function first, and then gave the people taking on the new roles the additional responsibility of examining those roles going forward. It was a year to eighteen months before we even thought about re-engineering.”

JPMorgan wanted to move client reference data, equity and fixed income trading accounts, security master and universal data such as calendars to the hubs, and started with the “easy ones” of each type, Serenita said. Currently it has migrated around 95 per cent of client data operations, some 70 per cent of security master, and up to 100 per cent of trading accounts and universal data.

While logic might suggest moving all functions to the lowest cost hub, he suggested that business continuity requirements demand a more balanced approach. “If you are trying to recover a function at a site, you can only do that if the function is similar or close to one that already resides there,” he said. “As part of the regular process we failover every month. When these functions resided in the business, they usually had a relatively local continuity plan. But what happens if the region is out? Another thing we got from migration was a fully balanced resiliency plan.”

Serenita also offered some words of caution on metrics and SLAs. “People take the idea of the SLA too far sometimes, and it gets a bad rep,” he said. “An SLA is just about trying to understand what the business requirement is, for example, what accuracy and what level of timeliness do we need? Does the business unit understand the cost factor of data delivered within three minutes or 15?”

Metrics should certainly be applied – “if you can’t measure, you can’t manage” – but need to evolve over time as the focus shifts from phase one of this process – migrate and stabilise – towards the later phases, where the emphasis should be on becoming content-centric, to make the operational hubs “the go-to experts – the first to get a call”, he said. “If you expand the business knowledge of the team, expand their roles and responsibilities to become content-focused, this leads to improved data quality.” This means firms can shift from applying metrics such as volume, turnaround time and process quality, to taking a more sophisticated approach and measuring factors like STP rates.
JPMorgan achieved savings in salary and benefits of 47 per cent. And data quality has improved. “How do I measure that? I get less phone calls,” said Serenita.

Related content

WEBINAR

Upcoming Webinar: Managing the transaction reporting landscape post Brexit: MiFID II, SFTR, EMIR

Date: 16 March 2021 Time: 10:00am ET / 3:00pm London / 4:00pm CET Duration: 50 minutes The transaction reporting landscape has, for many financial institutions, expanded considerably in size since the end of the UK’s Brexit transition period on 31 December 2020 and the resulting need for double reporting of some transactions to both EU...

BLOG

FINOS and Goldman Sachs Release Legend Open Source Data Management Platform

In pursuit of data collaboration and standardisation across capital markets, the Fintech Open Source Foundation (FINOS) and Goldman Sachs have released Legend, the bank’s flagship data management and data governance platform. The source code for five of the platform’s modules is available from FINOS. Goldman Sachs contributed its PURE logical modelling language and Alloy platform...

EVENT

RegTech Summit Virtual

The RegTech Summit Virtual is a global online event that will be held in May 2021 with an exceptional guest speaker line up of RegTech practitioners, regulators, start-ups and solution providers to collaborate and discuss innovative and effective approaches for building a better regulatory environment.

GUIDE

FRTB Special Report

FRTB is one of the most sweeping and transformative pieces of regulation to hit the financial markets in the last two decades. With the deadline confirmed as January 2022, this Special Report provides a detailed insight into exactly what the data requirements are for FRTB in its latest (and final) incarnation, and explores what needs to be done in order to meet these needs on a cost-effective and company-wide basis.