Executive Summary
The reference data problem continues to confound the financial services industry. High levels of data siloization, multiple external vendors and disparate formats and lack of standards have led to poor quality reference data. This has stymied trade processing automation efforts on the one hand while also putting pressure on operating costs because of highly manual error rectification and excessive duplication. Further, the market uncertainty and new regulation such as the Patriot Act are driving the urgency to accurately identify and manage risks with respect to clients and different counterparties. Organizations spend up to 10% of their IT budget on data management and still find themselves fighting data quality problems. The business case for reference data management could not be more compelling. The top 5 lessons from our reference data experience with several brokers, investment managers and custodians are as follows:

	A necessary and critical first step in reference data management is clean data. No amount of sophistication in your technology solution is going to make up for poor quality data. Continuing skepticism from the users about the integrity and quality of data will only result in a failed implementation. Get the data right, first and foremost
	While it is theoretically correct to aim for enterprise wide integration, it is practically more feasible to build your reference data solution step by step, prioritizing business applications and data types with the highest business impact, while keeping your approach and architecture open and flexible for enterprise wide integration
	Prioritize securities data and client/counterparty data for downstream business applications such as clearing & settlement systems and exposure management systems where the business impact of reference data will be the highest and where cost efficiencies will be realized rapidly
	Use a combination of build and buy and outsource to maximize your gains. Pure off the shelf products need heavy customization to fit with your organization. ASPs remain a question mark on the critical data sensitivity issue
	While the core application is best built to your specific requirements, there are potentially large savings to be realized from outsourcing the data and exception management operations to an offshore facility. Organizations employ up to 200 FTEs for these operations and are looking to save up to $10-12mn by outsourcing data management.

Reference data, rightly managed, presents a significant savings opportunity while also ensuring that your organization is equipped to manage risk and comply with regulatory requirements. Its time for every broker, investment manager and custodian to put reference data on their list of strategic technology initiatives.

The Challenge
Reference data, strictly defined, is the static information on securities and entities. In practice, the problem is often expanded to include market data and operating data central to risk management and management reporting.

Exhibit 1: Types of reference data

Securities data	Entities data	Market and operating data
Securities identifiers, description, sector codes	Counter party data - identifiers, BIC codes, account numbers, settlement details	Price data
Equity attributes - earnings ratios, dividend rates	Client data - account information, product usage, credit profile	Corporate actions notifications
Fixed income attributes - payment schedules, coupon rates, call/put features		Client portfolio and investment accounting data
Historical prices and index data		Position and AUM data
Currency and country codes, tax rates

Reference data is the life blood of the trade process and used through the trade processing cycle.

Exhibit 2: Usage of reference data across the trade process



Source: Capco

The financial services industry faces a fundamental yet complex problem in reference data management. Even while trade processing automation reaches new levels, inaccurate and inconsistent reference data remains the single largest stumbling block to internal STP. Equally significant for a profitability challenged industry, are the burgeoning costs of manually maintaining and updating data, error rectification and maintenance of duplicate and often redundant systems. Add to it, the risk of underestimated or unknown exposure to clients and counterparties.

From our experience with firms across the industry, several reasons have contributed to creating the reference data problem. The most significant of them is data siloization - the Tower Group Sept 2002 survey showed that firms have 37 systems on an average containing the same securities and counter party reference data with 8% of the firms having more than 150 systems. These are most often disparate systems, geographically distributed with no common formats or standards. Duplication of data across these disparate silos coupled with manual and inconsistent updates result in multiple representations for the same data elements making data quality a huge issue.

Adding to the problem are multiple external sources of market data again with little standardization between them - on an average data is purchased from more than 7 different vendors. Surprisingly this is an area where there is little automation - manual updates are still the norm. Eventually the dubious data quality leaves every downstream user application with no choice but to have its own data correction and maintenance operation leading to high manual operating costs.

In this scenario, while most firms are recognizing the need to act, they're caught wondering where to begin.

The Approach
Typically, questions faced by CIOs looking to start acting on reference data management are in the following areas:

Exhibit 3: CIO Questions

	Should I aim for an enterprise wide solution vs. a step by step approach?
	How do I prioritize reference data types and user applications and locations?
	How do I evaluate technology solutions from bespoke developers vs. off the shelf products vs. ASPs?
	How do I phase out my investment and expedite the savings and benefits?

From our experience with brokers and investment management firms, while it is theoretically correct to aim for enterprise wide integration, it is practically more feasible to get a buy in from one or more business units and build your solution step by step while keeping your approach and architecture open and flexible for enterprise wide integration. A very significant challenge firms face for enterprise wide initiatives is the need to centralize data administration. This is an organizational challenge with significant change implications for organizational structures, hierarchies and roles. This is best dealt with once minimum scale has been achieved and the benefits and returns are more visible across the organization.

We have found it useful to undertake the prioritization of reference data initiatives based on two factors - where is the business impact likely to be higher and where can I start realizing the cost efficiencies sooner.



On prioritizing between types of reference data, we believe securities data is the foundation of the trade process and critical to all business applications. We did find however, that banks and securities firms in Europe are focusing on improving their exposure and risk management practices and therefore according high priority to client and counterparty data. This will rapidly become the norm in the US as well, given the urgency created by Patriot Act and "know your customer" regulation. Investment management firms under heavy scrutiny from plan sponsors are prioritizing portfolio and performance data for more effective client and management reporting.

Build and buy and outsource
The evaluation of technology solutions appears complex on the face of it, but is actually quite simple if we look at different components of the solution. Each component of the solution could be bought off the shelf, built or outsourced and an optimal combination can be arrived at. Solutions available in the market, however, seem to be segmented traditionally by product, ASP and service offering as follows:

Exhibit 4: Comparison of technology solutions

Compared to these stand alone options, we believe there is an effective solution to be found through a combination of build and buy and outsource.

We believe that an optimal combination of build buy and outsource should be leveraged to realize maximum benefits as follows:

Build
Our case experiences with several firms lead us to believe that the reference data repository, the underlying data model and the integration with upstream sources and downstream user systems requires a very large element of customization specific to your firm's technology architecture, platforms and legacy systems. Most firms also consider the reference data repository to be the heart of the solution and a competitive differentiator which is worth building.

Buy
Some components such as the data dictionaries are industry standards to be complied with. Further, the appropriate middleware, integration & reporting tools and packages can be bought off the shelf and integrated into the solution.

Outsource
Error rectification and exception management are functions which firms are looking at outsourcing. The exception management system would entail the identification of incorrect data and rectification through a team of data analysts. Outsourcing this function delivers the twin objectives of cost rationalization and superior service. The 24*7 offshore delivery model has made it possible to deliver cleansed and validated data before the start of the trading day while also eliminating graveyard shifts for many an unwilling data analyst.

The systems integration approach to a reference data solution is clearly the right approach for building a robust solution at an effective cost.

Summary
We believe the flip side of the reference data management problem is the significant opportunity it presents to FSIs to cut costs and manage risk better. A solution covering different reference data types and multiple source systems and user applications can typically be delivered through a "build and buy and outsource" model with an investment of $5-10 mn. When it comes to benefits, firms can expect 15-20% reduction in trade processing failures. The cost savings come from various fronts. We have seen mid to large sized brokers and investment managers save upto $10 mn per annum from retiring multiple repositories and through lower manual processing and reduced operating and people costs. Consolidation of data vendors can shave off up to 25% of your data purchase costs (An average firm spends $3-5 mn per annum on purchasing reference data). Add to that the lowering of operational risk with accurate assessment and reporting of party wise exposure. That makes reference data management a very worthwhile candidate on your list of strategic technology initiatives.

This article is authored by Kavita Iyer a Consultant from the Solutions Design Group of Wipro Technologies. For more information send a mail to info@wipro.com