The Reference Data Hurdle: Managing Complexity Across the Insurance Enterprise
For managing insurance investment portfolios, you want a holistic view of risk, performance, and capital charges. That requires a unified language among securities, ratings, and benchmarks.
Part of this challenge is unique to insurance. Unlike other investment managers, insurers invest assets back into liabilities for products such as whole life policies, indexed annuities, property and casualty reserves. These have their own regulatory capital requirements and duration constraints.
Centrally managed reference data and a consistent taxonomy are essential for end-to-end views of individual portfolios. For example, for one portfolio, actuaries need to see how assets tie to products. Finance needs legal entity and jurisdictional splits. Investment teams managing external money have to pull credit quality for RFPs.
If they all look at the same positions through different lenses, they’re missing a common denominator. Each additional managed portfolio (either for the general account or for unaffiliated insurers) compounds that problem. Analysis and data rollups become increasingly cumbersome.
The industry is well aware of its challenges. An Oliver Wyman survey found 45% of insurers struggling with data timeliness and 39% with ongoing management of asset-level data.i Common underlying reference data, i.e., the identifiers, classifications, and attributes that define a security for pricing, risk, and reporting, is a critical piece for solving that challenge.
The lack of a common denominator
The fundamental problem is the differences between market data providers.
For example, Intercontinental Exchange (ICE) maintains nearly 300 classifications across asset classes, with products in granular buckets based on deal structure. Instruments like Z-bonds, planned amortization classes (PACs), and interest- or principal-only strips are separate in ICE, but another provider may roll them into a single mortgage-backed security.
Ratings create a similar disconnect. One agency may rate a security A- while another rates it BBB+. S&P A- is comparable to Moody’s A3, although not identical. These differences feed into insurers’ capital requirements, because ratings determine risk-based capital (RBC) requirements and capital charges. They map to 20 distinct National Association of Insurance Commissioners (NAIC) designation categories on a scale of one to six. Even minor differences create large capital consequences that can squeeze opportunities.
An insurer with multiple market data providers across desks eventually discovers that “the same thing” isn’t really the same across systems. Even when the CUSIP matches, one feed tags a tranche as a PAC and another rolls it up under a broader category. The same thing happens with historical and current ratings. Two teams can hold the same security and still compute different prices, risks, and rollups because their local definitions of the instrument have drifted.
“The General Investment Account (GIA) consists primarily of assets that support our insurance and retirement products. We organize the assets into smaller portfolios to better manage the assets relative to the liabilities. The nature of the product liabilities serves as the foundation for the investment policies that are developed for each portfolio.” — MassMutualii
The common workaround is normalizing this data in Excel. Teams export positions to a spreadsheet, apply local rules and their own mappings, and produce parallel versions of the truth. This tactic breaks down at the firm level because, eventually, someone has to pick a single answer to questions about instrument types or NAIC designations. Every rollup turns into a negotiation.
Large insurers face an additional challenge with reference data because they act as managers for other investors. The same people managing general account assets for liability matching often run external money for unaffiliated insurers or pension plans. One team needs duration-qualified instruments for claim payouts. Another needs credit quality data for RFPs. Differences between them tend to reduce reporting confidence.
What "good" looks like
Good reference data is a mastered layer that lets actuarial, finance, and investment teams ask different questions of the same positions without changing the underlying facts. Ultimately, it gives operators a stable foundation for daily decisions across every sleeve and makes it possible to give boards and mutual holders clear and defensible answers about consolidated exposures, credit quality, and capital-relevant classifications.
Normalizing ratings and asset taxonomy lets you support multiple lenses without creating multiple truths. Actuarial teams should be able to aggregate assets by product backing and sleeve. But finance needs the same holdings aligned with legal-entity structures and offshore exposure, with clean consolidation. And investment teams managing external mandates must have consistent credit quality and sector rollups for strategy reporting and RFP responses.
How to address the complexities
Addressing reference data complexity starts with introducing the missing common denominator. Because market data and benchmark providers categorize securities differently, firms have to develop a single, internal view for consistent exposure, risk, and performance reporting. From there, they should map each provider’s classifications and attributes into that standard with documented rules and owned exceptions.
Next, put central mastering at the center of the control model. A best-in-class setup maintains a security and instrument master alongside transactions, valuations, corporate actions, and the reference attributes that feed rollups. Keep the provider value and the normalized value together, then track effective dates so changes in ratings or classifications stay explainable over time.
Then address the operating cadence that keeps the model clean:
- Daily controls should check the completeness and validity of inbound files and surface schema drift early, with exceptions routed to an owned queue.
- Weekly governance should close open items and tighten mapping rules, with clear ownership for taxonomy, ratings history, NAIC mapping, and legal-entity tagging.
- Month-end controls should reconcile positions back to the centralized master before teams finalize risk, performance, and capital views.
Finally, protect the last mile and plan for the cases that create additional complexity. Spreadsheet mashing creates missed rows, combined securities, and changes that lack auditability. Keep downstream consumption tethered to the mastered layer to avoid this.
In addition, larger insurers that allocate across many unaffiliated managers also inherit timeliness differences, so tight ingestion, governance, and oversight become operational requirements rather than clean-up work.
What comes next
Reference data is infrastructure. It determines whether a portfolio behaves like one coherent book or an uneven collection. It also determines whether capital, risk, and performance views stay aligned as providers, ratings, and classifications evolve. This is ongoing work. The question is whether it is controlled work.
Senior leaders can pressure-test their posture by tying directly to capital outcomes and operational reality.
Insurers stay ahead by treating reference data as a standing operating discipline with defined standards, ownership, and cadence. Those that don’t keep paying for it in exceptions, tie-outs, and wasted leadership attention.
Authored By
Phillip Silitschanu
Phillip Silitschanu leads Arcesium's global digital asset commercial efforts as Senior Vice President, Digital Assets. Phillip is an expert and thought leader in the FinTech, blockchain, cryptocurrency, and digital assets space, known for his work as the research director leading IDC’s (Blackstone) global blockchain practice, and in various strategic roles within the financial services industry. He has authored and co-authored numerous whitepapers, reports, and books on these topics and is a recognized speaker and expert cited by major media outlets like the Financial Times and CNBC.
Share This post
