Corporate Actions Data Breaks After Vendor Migration
Northbridge Asset Management, a UK manager with £18bn AUM, migrated its corporate actions feed from Vendor A to Vendor B over a weekend. By Tuesday, 37 open positions across 12 FTSE 350 names showed entitlement mismatches between custody records and the investment book. Nine were dividend claims, three involved rights elections, and five had already been manually patched by operations. The vendor SLA promises 99.5% accuracy but does not define timeliness for complex events. Month-end NAV is six business days away, two data operations staff are on leave, and Vendor B's next release is in ten days. The COO must decide whether to revert to Vendor A, run both feeds in parallel, or accept manual remediation while B fixes mapping. Client reporting is due in four days and the regulator expects accurate records. The data governance committee meets tomorrow.
1. Identify the likely root causes and the immediate control you would put in place before the governance committee.
Read the worked answer
The pattern points to a mapping and identifier problem, not a single bad vendor file. Dividend claims, rights elections and custody breaks together suggest that Vendor B's event taxonomy, security identifiers and effective dates do not align with the firm's security master or custody records. The five manual patches are also a control weakness: they may have fixed the visible symptom without correcting the source mapping, and they create an audit trail gap.
The immediate control should be a documented exception bridge, not a free-for-all fix. Freeze ad hoc patches, require maker-checker for any manual entitlement adjustment, and open a single break log tied to position, event and custody reference. Run Vendor B alongside custody records for the 37 positions, prioritising the three rights elections because they have hard market deadlines. Do not rely on the vendor's accuracy SLA until timeliness for complex events is defined. The reasoning is that reverting mid-cycle creates a second migration risk, while accepting manual remediation alone leaves the firm unable to distinguish vendor error from internal mapping error.
2. Recommend an option to the COO and explain the trade-offs.
Read the worked answer
I would recommend controlled dual-running with a decision gate before month-end, rather than immediate reversion or open-ended manual remediation. Dual-running means Vendor B remains the target feed, but custody reconciliation and Vendor A data are used as an independent check for high-risk events until B's mapping is corrected and evidence shows a stable break rate.
The trade-offs are real. Dual-running costs licence fees and staff time when two operators are already on leave, and it adds complexity during NAV week. Reverting to Vendor A avoids short-term disruption but may not be possible at short notice, and it resets the migration without proving whether B can meet the ten-day release. Manual remediation is fastest for individual claims but does not scale and weakens control. The decision gate should be a daily scorecard: zero unexplained rights elections, fewer than three open dividend breaks, and all manual adjustments maker-checked. If the gate fails by month-end minus two days, escalate to a temporary reversion for the affected event types only.
3. Define three data quality metrics and the governance actions that should follow.
Read the worked answer
First, completeness: percentage of announced events captured by Vendor B against custody and exchange notices, measured daily. Second, timeliness: percentage of events received before the market deadline for elections and claims, not merely before NAV. Third, accuracy: reconciliation break rate between the investment book and custody, segmented by event type. A fourth useful measure is exception ageing, because a small number of old breaks is more dangerous than a larger number of same-day breaks.
Governance actions should assign a named data owner for corporate actions, document the security master and event mapping, and add vendor key performance indicators covering complex-event timeliness as well as accuracy. The committee should require a root-cause report for the five manual patches, approve the dual-run gate, and set a formal exit criterion. The reasoning is that data quality is not a single percentage; completeness, timeliness and accuracy fail differently and need separate controls. Client reporting should be signed off only after the rights elections are independently confirmed.