Lesson 6 of 6
Treasury testing and controls
9 min
Listen to this lesson
Treasury auditors arrive with a short, sharp list: who can trade, who can change bank details, who confirms deals, how market data is controlled, whether valuations and hedge accounting are right, and whether payments are approved. A treasury test plan should answer each of these with evidence before go-live, and again after every significant change.
The controls auditors test
| Control | Risk addressed | Test evidence |
|---|---|---|
| Segregation of front, back office and accounting | Unauthorized or fictitious deals | Role tests: dealer cannot settle or post; access review by Nadia |
| Deal release above limits | Deals beyond authority | Deal above threshold stops until a second person releases it |
| Confirmation matching | Deals not agreed with the counterparty | Matched and deliberately mismatched confirmations |
| Counterparty and dealing limits | Excessive exposure to one bank | Deal breaching a limit is flagged or blocked |
| Market data control | Wrong valuations | Rates reconciled with provider; entry separated from valuation |
| Bank account and payment approval | Payment fraud | Changed bank details need approval; payments above limit approved |
| Valuation and hedge accounting review | Misstated earnings | Hand-calculated expectations matched in 0L and 2L |
Treasury controls and the evidence that proves them.
A treasury test calendar
Treasury processes are time-driven: value dates, month-ends, maturities. Testing must simulate time. Contoso's test lead Priya planned treasury tests over two simulated month-ends, so that deals are created in one period, accrued and valued at its end, and mature or are valued again in the next. That is the only way to test accrual reversals, valuation resets and hedge effectiveness over consecutive periods.
Reconciliation
| Reconciliation | Compare | Frequency |
|---|---|---|
| Treasury positions to G/L | Position values per valuation area with the G/L accounts in each ledger | Each month-end |
| Deals to counterparties | Open deals with bank statements of outstanding deals | Monthly or quarterly |
| Internal accounts to intercompany | In-house bank internal accounts with intercompany balances | Daily or monthly |
| Cash forecast to actual | Treasury flows forecast with bank statements on value date | Daily |
Treasury reconciliations that support the close.
What would you do?
Two weeks before go-live, a treasury user asks for permission to change counterparty bank details directly in production "to save time", since deals with a new bank start next week. What should happen?
What goes wrong in testing
- Tests run with one super-user, so every segregation control is untested.
- Only one month-end is simulated, so reversals and resets are never seen.
- Expected values are not prepared before the run, so testers confirm whatever the system produced.
- Converted open deals from ECC are not revalued and reconciled in the first mock, so opening positions are unproven.
The last point matters for Contoso's conversion: open treasury deals in EP1 are converted with the system. Their positions, accrued interest and valuation history must reconcile with the general ledger after conversion, per valuation area, before anyone values them again. Add a treasury position reconciliation to the finance reconciliations Dan signs at go/no-go 2 on Sunday at 14:00.
- Can each role do only its own steps?No → fix roles before any other treasury test · Yes → next
- Does every product type pass its lifecycle with hand-calculated values?No → fix configuration and retest · Yes → next
- Do two simulated month-ends reconcile in 0L and 2L?No → investigate valuation areas and accounts · Yes → next
- Do converted open positions reconcile with the G/L?No → block treasury go-live · Yes → ready for Dan's sign-off
After go-live, the same controls run every month, and the test pack becomes the regression pack. When SAP releases a support package or a new feature pack, or Contoso adds a product type, the affected lifecycle tests and the two-month-end valuation tests run again in the quality system before the change reaches production.
Evidence that survives the audit
Treasury evidence must let an auditor re-perform the test months later. For each lifecycle test, keep the deal number, the users who performed each step, the confirmation documents, the hand calculation and the posted documents per ledger. For valuation tests, keep the market data used, with its date and source. For access tests, keep the role assignments as of the test date and the access review sign-off. Screenshots alone rarely suffice; document numbers and reports with selection criteria do.
- Auditors test segregation, deal release, confirmations, limits, market data, payment approval and valuation.
- Simulate at least two month-ends to test accruals, resets and effectiveness.
- Reconcile treasury positions to the G/L per valuation area every month-end.
- Converted open deals must reconcile before they are valued again.
Check your understanding · not graded
Lesson check
1. Why simulate two month-ends in treasury testing?
2. Which control addresses fictitious deals?
3. What must happen to converted open deals before revaluation?
4. A user wants to change bank details directly to meet a deadline. What is right?