The Core Dimensions of Data Quality in Securities Lending

On an agency lending trading desk, data is more than numbers on a screen — it’s the foundation for trust between counterparties, operational efficiency, and regulatory compliance. When a trader agrees to lend securities, the desk and operations book the trade, risk teams monitor exposure, and reporting teams file to regulators. Each step relies on data being correct, consistent, and available. If the data fails, so does the business.

This is where data quality dimensions come in. They give us a lens to assess whether the data fueling our lending programs is truly “fit for purpose.” There list of dimensions that we want to explore are eight in number:

1. Validity

2. Completeness

3. Consistency

4. Integrity & Coherence

5. Timeliness

6. Reasonableness

7. Uniqueness

8. Accuracy

As a business case let us walk through how each dimension plays out in the real world of securities finance. We hope this helps our readers understand the practicality of applying concepts in data in their projects through our examples.

Validity: Keeping data within the rules of the game

Every trading desk works within guardrails. For example, ISINs must be 12-character alphanumeric codes, loan rates must sit within realistic ranges (0–100%), and collateral must carry valid ISO currency codes like USD or GBP. When values fall outside these boundaries — say, a loan rate of 500% or a non-existent “QQQ” currency code — systems downstream may break, or worse, miscalculate exposure. Validity ensures every data point “makes sense” within the rules of the market.

Completeness: No gaps, no blind spots

Imagine running an SFTR submission without a UTI or LEI — the report is instantly rejected. Or picture an open loan record missing collateral details; the desk now has unreported risk exposure. Completeness lives at different levels:

• Column level: Fields like currency codes or action types can never be blank.

• Record level: Open loans need collateral type, value, and haircut populated.

• Conditional: Cash-collateralized trades require an interest reinvestment rate; non-cash trades must leave it blank.

• Dataset level: Daily loan files from the agent lender must include every active trade, no omissions. Without completeness, decisions are made in the dark.

Consistency: Speaking the same language across systems

Inconsistent data creates noise. If the loan currency is USD but collateral currency shows EUR — without an FX agreement in place — red flags rise. If positions in the agent lender’s file don’t match those in the custodian system, reconciliations stall.Consistency means:

• Records agree internally (loan and collateral currencies line up).

• Trades agree across datasets (custodian vs. lender).

• Reports stay stable over time (T+1 submissions don’t change retrospectively unless valid modifications occur).

Consistency keeps the story straight.

Integrity & Coherence: Keeping relationships intact

Think of fund hierarchies: if a trade is booked to a sub-fund, it must roll up to the correct umbrella fund. Break that relationship, and aggregated reporting falls apart.Integrity ensures references aren’t orphaned (no “child” without a “parent”), while coherence keeps values logically aligned. Duplicate trades or mismatched hierarchies distort both exposure and client reporting.

Timeliness: Data that arrives too late is data lost

Securities finance is fast-moving. Trades must be booked as soon as deals are struck, recalls must be issued the moment a beneficial owner asks, and collateral valuations should refresh intraday as markets shift. Timeliness has layers:

• Static data like LEIs may stay valid for years but must be reviewed annually.

• Slowly changing data like counterparty credit ratings should be refreshed monthly.

• Volatile data like collateral values demand intraday updates. If timeliness slips, decisions lag — and risk grows.

Reasonableness: Does the data “smell right”?

A haircut of 150%? (typically bonds 2-5%, equities 20%+), A settlement fail rate jumping from 2% to 20%? Reasonableness tests flag these anomalies. By comparing data against thresholds, benchmarks, and business norms, reasonableness helps spot the outliers before they become costly errors.

Uniqueness: One trade, one record

Every trade ID must be unique. Duplicate trades double-count exposure, revenue, and risk. Similarly, each security should map to a single ISIN — no duplicate identifiers floating around. When uniqueness breaks down, reconciliations pile up and clients start calling.

Accuracy: Reality reflected in data

Ultimately, accuracy is about truth. Do the economics of the trade in the system — loan amount, rebate rate, collateral value — match what the counterparty agreed in the confirmation? Do legal entity names and LEIs match official registers? Are market prices aligned with trusted sources? Accuracy is tested through reconciliation, sampling, and verification against golden sources. It is the dimension most directly tied to client trust.

Bringing it together: From rules to reality

Each of these dimensions works like a lens, focusing on a different aspect of whether data is truly “fit for purpose.” Business rules tell us how data should be collected and structured. Data quality dimensions tell us whether it’s usable for lending, reporting, and risk management.

For a securities lending desk, the story is simple: when data quality fails, risk, revenue, and reputation are on the line. When it succeeds, clients gain confidence, operations flow smoothly, and the business can grow.