Reference data governance is the framework through which a financial institution manages the descriptive data that its transactions depend on. Reference data identifies and describes the things being traded and the parties involved: securities and their terms, issuers, counterparties and clients, accounts, currencies, calendars, markets and settlement instructions. Governance assigns responsibility for each of these data sets, defines the standards they must meet, and puts controls in place to check that they do.
The subject is a specialised part of enterprise data governance. It matters in finance because the same reference items are used by many systems at once, in trading, risk, settlement, accounting and regulatory reporting, so one wrong attribute can cause failed trades, incorrect valuations and misreported positions throughout a firm. Regulators have made data quality a supervisory concern in its own right.
What reference data covers
The main domains are instrument data and party data. Instrument reference data includes identifiers such as the ISIN, classification, issue terms, coupon and maturity details, and trading and settlement conventions, and is commonly held in a security master file. Party data describes legal entities, their identifiers such as the legal entity identifier, their ownership hierarchies and their regulatory classifications; the issuer side is covered under issuer reference data.
Other domains include standing settlement instructions, corporate action data, pricing sources, holiday calendars, and internal structures such as books, accounts and product hierarchies. Much of this is bought from data vendors, exchanges and numbering agencies and then enriched internally.
Components of a governance framework
Although organisations differ, an effective framework usually contains the following:
- Ownership. A named senior owner is accountable for each data domain, supported by data stewards who manage definitions and quality day to day. A data governance council or committee resolves conflicts and sets priorities.
- Standards and definitions. A business glossary and data dictionary state what each attribute means, its permitted values and its format, so that the same field is understood identically across systems.
- Authoritative sources. For each attribute, one source is designated as authoritative and all consuming systems take the value from it. This is the principle of golden source data.
- Quality rules and monitoring. Automated checks test completeness, accuracy, timeliness, consistency and uniqueness, with results measured against thresholds and reported.
- Change control. Additions and amendments follow a controlled process, typically with separation between the person who enters a change and the person who approves it, and with an audit trail.
- Issue management. Errors are logged, traced to their cause and corrected at source instead of being patched in downstream systems.
- Lineage. Documentation records where each data element originates and how it flows and is transformed on its way to reports.
Regulatory drivers
The principal international reference is the Basel Committee's Principles for effective risk data aggregation and risk reporting, published in 2013 and widely known by the document number BCBS 239. Written after the financial crisis, when many banks could not total their exposures to failing counterparties quickly, it requires systemically important banks to have strong governance over risk data, a data architecture that supports aggregation, and the ability to produce accurate reports in normal and stressed conditions. Supervisors have repeatedly reported that full compliance remains incomplete, and the European Central Bank has made data aggregation a supervisory priority.
Transaction and trade reporting regimes depend directly on reference data. Reports under MiFID II and derivatives rules must carry valid instrument and entity identifiers and product classifications, and errors lead to rejected reports and, in a number of cases, regulatory fines. Client data supports know your customer and sanctions screening obligations, and accounting and prudential reports depend on correct classification of instruments and counterparties.
Operating models
Many firms centralise reference data in a dedicated function that sources, validates and distributes it to consuming systems, replacing arrangements in which each business line maintained its own copy. Centralisation reduces duplication and inconsistency, though it requires agreement on common definitions that business lines may resist. Some firms outsource maintenance to managed service providers or use industry utilities for shared data such as legal entity records.
Common identifiers are a precondition. Standards maintained through the International Organization for Standardization for securities, entities, classification and currencies allow data from different vendors and systems to be matched. Where a firm still relies on proprietary or vendor-specific codes, it must maintain cross-reference tables, which are themselves reference data needing governance. A related discipline is vendor master data in the finance function.
Common weaknesses
Typical problems include unclear ownership, especially for data that crosses business lines; multiple inconsistent copies of the same data; manual corrections made in downstream systems that never reach the source; late set-up of new instruments, which delays trading or settlement; and weak controls over vendor feeds. Stale security static data is a frequent cause of settlement failure and of errors in corporate action processing.
Governance programmes are also difficult to sustain. Their benefits appear as the absence of errors, which is hard to demonstrate, while their costs are visible, so they compete poorly for budget unless a regulatory finding or an operational loss forces the issue. Firms that succeed tend to measure data quality in terms the business recognises, such as failed trades, reporting rejections and manual effort, and to tie ownership to senior management accountability.