Does Covenant Monitoring Software Calculate the Covenant or Just Track the Deadline?

Most of it tracks. It stores the ratio the borrower certified, records that the certificate arrived, and turns the diary item green. Software that *calculates* starts from the audited accounts, rebuilds the ratio to the facility agreement's own definitions, and produces a number the borrower did not supply. Ask which one you are buying.

Key facts

• Recomputation is an extraction problem first. YuSight measures 95.2% extraction accuracy against a manual benchmark, every figure traced to source document and page. A system that cannot read the lease note cannot include finance-lease principal in fixed charges, and the worked example below turns entirely on that line.
• Borrowers now disclose their covenants in their own notes. Since annual periods beginning on or after 1 January 2024, IAS 1 requires that covenants complied with after the reporting date "do not affect the classification of debt as current or non-current at the reporting date" and that companies instead "disclose information about these covenants in the notes to the financial statements" (IFRS Foundation, October 2022). That note is a free cross-check on what the borrower thinks the covenant says.
• Monitoring is a supervisory expectation across the loan's whole life, not just at origination. The EBA Guidelines on loan origination and monitoring (EBA/GL/2020/06, published 28 May 2020, applying from 30 June 2021) "specify the internal governance arrangements for granting and monitoring of credit facilities throughout their lifecycle" (EBA).
• Manual computation is not neutral in supervisory eyes. BCBS 239 Principle 3 requires risk data to be "aggregated on a largely automated basis so as to minimise the probability of errors," with "effective mitigants in place" wherever a bank relies on spreadsheets and desktop applications (BIS, January 2013).
• In the worked case below the certified ratio is 1.15x and the recomputed ratio is 1.01x — a 12.1% shortfall against the covenant, from three definitional lines, on a compliant-looking certificate.

What does "calculate" actually mean here?

There are four distinct things vendors call calculation, and only the top two are what the word implies.

Tier 2 is where most disappointment lives, because it looks like calculation. Something divides. A ratio appears. But if EBITDA came off the certificate and went into a form field, the platform has performed long division on the borrower's assertion. The category-level version of this distinction is in covenant tracker vs covenant monitoring software; this page is about the single question that separates tier 2 from tier 3.

Where do the input figures come from?

Four possible sources, in descending order of how much you should trust them and ascending order of how often they are used.

1. Your own core banking and loan systems. Drawn balances, interest charged, scheduled amortisation, facility limits. This is the most reliable data in the entire test and the most frequently ignored, because it lives in a different system from the covenant module.
2. The borrower's statutory accounts. Audited, with notes. The only source that supports finance-lease principal, related-party balances, restricted cash and the disclosure of covenant terms now required under IAS 1.
3. Management accounts. Timely, unaudited, and the basis on which most quarterly maintenance covenants are actually tested.
4. The compliance certificate. A contractual representation, and a valuable one — someone signed it. It is evidence of what the borrower claims, not evidence of the ratio.

A tier 3 system uses sources 1 to 3 and treats source 4 as the thing being checked. A tier 1 or 2 system uses source 4 and calls it an input.

A worked case: certified 1.15x, recomputed 1.01x

A mid-market industrial borrower. $85m term facility, quarterly maintenance covenant: Fixed Charge Coverage Ratio ≥ 1.15x, tested on a rolling twelve-month basis. Certificate due within 45 days of quarter end. Figures in $'000. The facility terms and figures are illustrative.

What the certificate says:

It rounds to the threshold exactly. That alone should route the file to a human.

What recomputation from the accounts produces — three differences, each traceable to a clause:

• Non-recurring gain. EBITDA includes a 1,100 gain on disposal of a facility. The agreement's EBITDA definition excludes gains on asset disposals. Adjusted EBITDA = 24,800 − 1,100 = 23,700.
• Finance-lease principal. The Fixed Charges definition includes principal on capital leases. The lease note discloses 900 paid in the period; the certificate omits it.
• Rent understated. The lease note shows rent expense of 2,900, not 2,600 — a new lease commenced mid-period and the certificate used the prior-year run rate.

Recomputed numerator = 23,700 − 3,200 − 1,900 = 18,600 Recomputed fixed charges = 6,400 + 8,200 + 900 + 2,900 = 18,400 Recomputed FCCR = 18,600 ÷ 18,400 = 1.0109 → 1.01x

Breach. Required numerator at 1.15x = 1.15 × 18,400 = 21,160. Actual 18,600. Shortfall 2,560, or 12.1% of what the covenant demands.

Every one of the three differences was disclosed in a note the borrower filed. None of them required a phone call.

What is that difference worth?

This is the part that decides the budget question, and it is not the licence fee.

• A waiver has a price. On an $85m facility, a waiver fee of 25bp of commitments is $212,500, one-off. A 50bp margin uplift on $85m drawn is $425,000 a year for as long as it runs. Neither is available to a lender that recorded "compliant" and closed the item.
• A breach buys consent rights. With a breach on the file, a proposed $4m distribution to shareholders requires your agreement. Without it, the distribution is simply made.
• Timing is the real asset. Recomputation catches this at the Q2 test. Certificate acceptance catches it — if at all — at the annual audit, typically nine to twelve months later, by which point the $4m has left and the borrower's options have narrowed.

Set against those three, the question "does it calculate?" stops being a product-feature question. The definitional battleground itself is mapped in covenant testing for DSCR and leverage.

How are covenant definitions configured per facility?

Badly, in most implementations, and this is the failure mode worth naming because it converts tier 3 software back into tier 1.

A workable configuration has three layers:

1. A formula library — market-standard FCCR, DSCR, net leverage, interest cover — as a starting point only.
2. Per-facility parameters bound to the clause that sets them: the add-back cap and its basis, the pro-forma treatment of acquisitions, cash-netting limits, whether frozen GAAP applies and from what date, the inclusion list for fixed charges, the rounding convention, and the cure mechanics.
3. A clause citation on every parameter — clause number, page, executed version — so a disputed input resolves to text rather than to whoever configured it.

Skip layer 2 and you have a platform producing confident market-standard ratios for facilities that were not drafted to market standard. That is worse than a tracker, because a wrong number gets believed. The full testing cycle these definitions plug into is set out in covenant monitoring in commercial lending.

Three probes that settle it in one demo

• "Show me a facility where the certified ratio and your ratio disagree." Not a screenshot — a live record. A tier 1 or 2 product has no such record, because it has only one number.
• "Where did the finance-lease principal come from?" Click it. It should open the lease note at the right page. If it opens a form field, an analyst typed it.
• "Change the add-back cap on this one facility to 10% of EBITDA and re-run last quarter." Tier 3 re-runs and shows the movement. Tier 2 asks you to re-key.

FAQ

Does covenant monitoring software calculate covenants?

Some does. A lot of it stores the ratio the borrower certified and manages the calendar around it, which is genuinely useful but is not calculation. The test is whether the tool can produce a ratio the borrower never sent you.

Where do the input figures come from?

In a system that really computes: your own loan system for balances and scheduled amortisation, and the borrower's statutory or management accounts for everything else, with the compliance certificate held to one side as the thing being checked. If the certificate is the input, you are reading the borrower's arithmetic back to yourself.

How are covenant definitions configured per facility?

You start from a formula library and then override it facility by facility — add-back caps, the fixed charge inclusion list, pro-forma treatment, frozen GAAP, rounding — with each parameter tied back to the clause and page that sets it. The configuration work is the product. Software that ships with one market-standard formula for four hundred differently-drafted facilities has not been implemented, only installed.

What if the borrower does not send the underlying accounts, only the certificate?

Then you can only monitor at tier 1, and you should say so in the file rather than let the green status imply otherwise. It is also a drafting point for the next facility: make delivery of the accounts a condition of the certificate being treated as delivered.

Does the software decide whether a breach has occurred?

No, and it should not claim to. It produces a recomputed ratio, an explained difference against the certified one, and a proposed determination. A named human accepts or overrides it, and that decision is the record — see designing the analyst override workflow.

Do the IAS 1 covenant disclosures actually help?

Yes, more than people expect. The notes now carry the borrower's own description of its covenants, which you can read against your executed agreement. Where the two differ, you have found a misunderstanding before the test date rather than after it.

Is a rounded certified ratio a red flag on its own?

A certified ratio that lands exactly on the threshold is not proof of anything, but it is the cheapest routing rule you can implement. Recompute every certificate that comes in within two decimal places of its covenant, whatever else you sample.

Conclusion

Three things to take away:

1. "Calculates" is a four-tier word. Establish which tier the product occupies before the commercial conversation, using the three demo probes above.
2. The certificate is evidence of a claim, not an input. Any architecture that treats it as an input has already decided that the borrower does the arithmetic.
3. The value is in the timing and the consent rights, not in the ratio itself. A breach found at the quarterly test is worth a waiver fee, a margin ratchet and a veto over a distribution. The same breach found at the audit is worth a conversation.

YuSight's Covenant Monitoring spreads the underlying financials at 95.2% extraction accuracy validated against a manual benchmark, recomputes each tested ratio to the facility agreement's own definitions, and shows the certified figure beside the recomputed one with every input cited to its source page.

See covenant testing run automatically — book a live demo.