Data Management, Data Quality and Governance
What this domain tests
The exam tests whether you can define data domains, apply the standard data-quality dimensions, explain master data and lineage, and describe how governance assigns accountability. It also expects awareness of BCBS 239 for risk data aggregation and GDPR for personal data.
Core rules and concepts
- Data domains: reference data (instrument, counterparty, calendar, price), market data, transactional data (orders, trades, positions), derived data (valuations, exposures) and client data.
- Quality dimensions: accuracy, completeness, timeliness, consistency, validity and uniqueness. Each needs a measurable rule, a threshold and an owner.
- Golden source / master data management: one authoritative record per entity, with controlled distribution to downstream systems. A golden copy is a practical target; a single source of truth across a large firm rarely exists.
- Lineage: the documented path from source to report, including every transformation. Without lineage you cannot assess the impact of a data error.
- Governance: data owners (accountable), stewards (operational), policies, data dictionary, quality dashboards, exception queues and escalation.
- BCBS 239: principles for risk data aggregation and risk reporting - governance, data architecture, accuracy and integrity, completeness, timeliness, adaptability, and supervisory review.
Worked example
A bond's maturity date in the security master is 2031 instead of 2030. The risk system computes modified duration from the wrong date, and fund accounting accrues coupon beyond legal maturity, misstating NAV. Reasoning: trace lineage from the vendor feed to the security master to risk and accounting. The root cause is that no validation rule compared the date against a second source or the prospectus. The fix is not a manual correction alone: add a data-quality rule (maturity after issue date; cross-source tolerance), require four-eyes approval for static-data changes, route exceptions to a monitored queue, and record the lineage so impact analysis is possible next time.
Traps that cost marks
Treating data quality as an IT problem when ownership sits with the business; claiming a single source of truth where multiple systems hold conflicting records; choosing timeliness over accuracy without documenting the trade-off; assuming vendor data is clean; ignoring end-user computing - spreadsheets outside governed systems are a classic exam trap and a real control gap.