This module teaches the lending lifecycle from zero, but goes deep enough for a professional
It assumes no prior knowledge of credit or loan operations and builds every idea from its foundations. Yet it goes deep enough to be useful to a credit analyst, a loan-operations manager, a risk-data architect or a consultant running a lending transformation.
Wherever a concept is technical, we stop and explain it in plain language before using it again. Wherever a term is jargon, we define it once, clearly, and then rely on it. French terms are given where the French market has its own vocabulary, and European regulation is used as the reference throughout.
Two threads run through everything: the business and the technology
The business thread covers how a bank decides whether to lend, how it shapes and prices the loan, what it does while the loan is alive, and how it gets its money back when things go wrong.
The technology thread covers the systems that carry a loan through those stages — credit workflow, rating engines, limit systems, loan servicing platforms, covenant trackers, provisioning engines and workout case management — and the data that must pass cleanly from one to the next.
In lending, the two threads meet in a single object: the facility record. A loan that was analysed well but booked wrongly accrues the wrong interest; a covenant that was negotiated well but never captured as data is never tested; a deterioration that was visible in the cash flows but not linked to the rating is provisioned too late. The lifecycle is only as strong as its weakest hand-off.
Where this module sits in the Corporate Banking track
This is the second module of the track. The first, on coverage, explained how the bank wins and owns a client. This one explains what happens to a loan from the moment coverage brings in an opportunity until the facility is repaid, refinanced or recovered. It factorises eight stages that every financing product shares, so that the product modules that follow — corporate and leveraged finance, restructuring, distribution, energy and real assets, agency, trade and cash management — can concentrate on what makes each product different.
The boundary is simple. This module explains the process: the steps, the controls, the systems and the data. The product modules explain the economics, markets and actors of each business line. Where a topic belongs mainly to a product module — syndication strategy, agency mechanics, restructuring advisory — this module introduces it and points forward.
How to read this module
Read it as a narrative, not a reference card
Each part assumes you have read the ones before it. If you are short on time, read the section titles and the "In a nutshell" boxes first, then come back for the detail. The titles are written to carry the argument on their own.
Each coloured box plays a fixed role
Key insight distils the one idea from a section worth remembering. Going deeper adds nuance for readers who already have the fundamentals, including the French and European specifics. IT & data lens connects the business step to the systems and data beneath it. Worked example grounds an abstract mechanic in concrete numbers. Watch out flags a common misunderstanding or a place where loans fail. In a nutshell closes a section with a short recap. Common misconception names a plausible wrong answer and dismantles it. Practice sets out calculations and judgement calls to attempt yourself. Explain it back asks you to teach the idea to someone outside finance.
One running case follows a loan from birth to resolution
The worked examples follow Helion Industries, the fictional Lyon-based aerospace and energy components group introduced in the coverage module. Helion has revenue of about 2.8 billion euros, EBITDA of about 420 million euros and net debt of about 900 million euros. In this module it asks the bank for a financing package to support the acquisition of a US competitor: a 150 million euro five-year amortising term loan and a 100 million euro revolving credit facility, both bilateral.
You will watch that package being analysed, rated, structured, approved, documented, booked and serviced. Then, eighteen months later, an aerospace supply-chain shock cuts Helion's earnings, a covenant comes close to breaking, the loan migrates between accounting stages, and the bank has to decide how to respond. Every number attached to Helion is invented to teach a mechanic.
The module is built to be worked, not only read
Practice boxes contain the calculations and judgement calls the credit, operations and workout teams actually make: a leverage ratio from raw accounts, an expected loss, an interest accrual, a commitment fee, a covenant headroom, a choice between restructuring and enforcement. Attempt them before looking anything up.
Explain it back boxes ask you to teach the idea to a specific person who does not work in finance. If you cannot do it without the module's vocabulary, you have not finished learning the section.
Common misconception boxes name the plausible wrong answer directly, because most of them are believed by people who work in the industry.
Check your understanding boxes close each part with recall questions whose answers are collected in the Answer Key. Retrieving an answer from memory is what fixes it.
A suggested path through the document
Four sessions work well. The first covers Part 1, which sets out the lifecycle and the credit-risk vocabulary that runs through it. The second and third cover Part 2, the eight stages, which is the heart of the module: stages one to four, from origination to approval, in one session, and stages five to eight, from booking to workout, in the next. The fourth covers Parts 3 to 5 — actors, systems and transformation — together with the summary and the cheat sheet, and is the right moment to return to the practice boxes.
Readers who already work in credit can treat Part 1 as revision, with one exception: the section on how the same three risk parameters feed both capital and provisions is new to many practitioners who see only one side.
What you will master
You will be able to walk a loan through all eight stages
You will follow a facility from origination through credit analysis and rating, structuring, approval, execution and booking, servicing, monitoring, and workout or exit, and you will know what each stage produces and who owns it.
You will speak the language of credit risk
You will understand probability of default, loss given default and exposure at default, how they combine into expected loss, how a rating maps to a probability, and why the same parameters drive both regulatory capital and accounting provisions.
You will understand how a loan is shaped and approved
You will know the levers a structurer pulls — amount, tenor, amortisation, pricing, security, covenants — and how a credit decision is routed through delegated authorities and committees to a conditional sanction.
You will be able to trace the operational life of a loan
You will know how a signed agreement becomes a booked facility, how interest and fees are calculated and collected, how drawdowns, rollovers and prepayments work, and how the money and the records are reconciled.
You will recognise deterioration early and know the responses
You will know how covenants, early-warning signals and re-rating detect trouble, how IFRS 9 staging moves a loan from twelve-month to lifetime provisions, what non-performing and forbearance mean, and how a workout team chooses between waiting, restructuring, selling and enforcing.
You will be able to map the systems and data front to back
You will follow the facility record through the credit workflow, the limit system, the loan servicing platform, the covenant tracker, the early-warning and provisioning engines and the general ledger, and understand why data re-keyed at any seam becomes a control failure later.
1Context and the Big Picture
Before following a loan stage by stage, you need the map and the vocabulary. This part defines what a corporate credit facility is, lays out the eight stages of its life, and introduces the three numbers — probability of default, loss given default and exposure at default — that every stage ultimately serves.
A loan is a promise with two ways out
Lending means handing over money against a promise
When a bank lends to a company, it hands over money today in exchange for a promise: the company will pay interest along the way and return the principal at agreed dates. Everything the bank does before lending is an attempt to judge whether that promise will be kept. Everything it does afterwards is an attempt to keep checking, and to protect itself if the answer changes.
The first way out is the borrower's cash flow
Bankers talk about two "ways out" of a loan. The first way out is the borrower's own operating cash flow: the company runs its business, generates cash, and uses it to pay interest and repay principal. A healthy loan is repaid entirely from the first way out, and the bank never needs anything else. Credit analysis is mostly the study of the first way out.
The second way out is what the bank can claim if the first fails
The second way out is what the bank can recover if the cash flow fails: collateral it can sell, guarantees from a parent or a third party, its ranking in the queue of creditors in an insolvency. The second way out does not make a bad loan good; it limits how much is lost when a loan goes bad. A banker who lends because the collateral is good, rather than because the business can repay, is lending on the wrong way out.
The lifecycle has eight stages in three phases
Deciding, putting in place, living with it
A corporate credit facility passes through eight stages, which group naturally into three phases.
The first phase is deciding. Coverage originates the opportunity. Credit analysts analyse the borrower and assign a rating. The deal team structures the facility: its amount, tenor, repayment, pricing, security and covenants. An authorised person or committee approves it, usually with conditions.
The second phase is putting it in place. Lawyers document the facility, conditions are satisfied, and operations book it in the loan system and activate the credit limit. This is execution and booking.
The third phase is living with it, and it lasts for years. Operations services the loan: drawdowns, interest, fees, repayments and amendments. Risk and coverage monitor it: covenants, early-warning signals, annual reviews and provisions. And eventually the loan exits — repaid, refinanced or, if it has gone wrong, worked out and recovered.
| Stage | What it produces | Main owner | Core system |
|---|---|---|---|
| 1. Origination | A qualified opportunity and a client need | Coverage | CRM |
| 2. Credit analysis and rating | Spread financials, a risk assessment, a rating and a PD | Credit analysis | Spreading tool, rating engine |
| 3. Structuring | A term sheet: amount, tenor, pricing, security, covenants | Product and structuring | Pricing and deal tools |
| 4. Approval | A conditional credit sanction and an approved limit | Credit officers and committees | Credit workflow (loan origination system) |
| 5. Execution and booking | Signed documents, satisfied conditions, a booked facility, an active limit | Legal, loan operations | Document management, loan servicing system, limit system |
| 6. Servicing | Drawdowns, interest and fee collection, repayments, amendments | Loan operations, agency | Loan servicing system, payments |
| 7. Monitoring | Covenant tests, early-warning signals, reviews, IFRS 9 staging and provisions | Coverage, credit risk, finance | Covenant tracker, early-warning system, ECL engine |
| 8. Workout and exit | Repayment or refinancing; or watchlist, restructuring, recovery, write-off | Coverage; workout unit | Case management, collateral and recovery systems |
The stages are not a straight line
The lifecycle looks sequential, but it loops. Analysis and structuring iterate: the structure changes when the analysis reveals a weakness, and the analysis is redone when the structure changes. Monitoring loops back to analysis at every annual review, when the borrower is re-rated and the facility re-approved. And any facility can jump from monitoring to workout when something breaks, and occasionally back again when the borrower recovers.
A handful of products cover most corporate lending
The standard building blocks
The product modules explore each financing product in depth. For the lifecycle, five building blocks are enough.
| Product | What it is | Typical use | How it is repaid |
|---|---|---|---|
| Revolving credit facility (RCF) | A committed line the borrower can draw, repay and redraw up to a limit during its life | General corporate purposes, liquidity back-up, working capital | Each drawing is repaid at the end of its interest period and can be redrawn; the line expires at maturity |
| Term loan | A loan drawn once or in a few tranches early in its life | Capital expenditure, acquisitions, refinancing | Amortising (instalments) or bullet (all at maturity) |
| Bridge loan | A short-term loan to be refinanced by a bond, equity or asset sale | Acquisition finance pending permanent funding | From the refinancing proceeds |
| Guarantee and letter-of-credit lines | Contingent commitments in which the bank promises to pay a third party if the client does not | Bid and performance bonds, trade | Not drawn unless called; then becomes a loan |
| Overdraft and uncommitted lines | Lines the bank can cancel or refuse to fund at any time | Short-term flexibility | On demand |
Committed versus uncommitted
The distinction that matters most for the lifecycle is between committed and uncommitted facilities. A committed facility obliges the bank to lend when the borrower asks, as long as the conditions in the agreement are met, even if the bank would prefer not to. An uncommitted line gives the bank discretion. Committed facilities are worth more to the borrower and cost more to the bank, because the bank must hold capital against the undrawn part and must have the funding ready.
Bilateral, club and syndicated
A facility provided by one bank is bilateral. One provided by a small group of banks on identical terms, usually from the borrower's panel, is a club deal. A large facility arranged by one or more lead banks and distributed to many lenders is syndicated, with an agent bank administering it on everyone's behalf. The lifecycle stages are the same in all three; what changes is the number of parties, the documentation and the data. The Distribution and Agency modules develop syndication in full.
Credit risk can be described with three numbers
Probability of default
The probability of default, or PD, is the likelihood that the borrower will default within a given horizon, usually one year. Default has a precise regulatory meaning that Part 2 sets out; for now, it means the borrower fails to pay, or is judged unlikely to pay, what it owes. A strong investment-grade company might have a PD of 0.05 per cent; a weak leveraged borrower 5 per cent or more. The PD is a property of the borrower, not of the loan.
Loss given default
The loss given default, or LGD, is the share of the exposure the bank expects to lose if the borrower defaults, after recoveries from collateral, guarantees and the insolvency process, and after the costs of recovery and the time value of money. An unsecured senior loan to a corporate might carry an LGD of 40 to 45 per cent; a loan well secured on real estate perhaps 20 per cent; a subordinated loan much more. The LGD is a property of the facility — of its security, its ranking and its documentation.
Exposure at default
The exposure at default, or EAD, is the amount the bank expects to be owed at the moment of default. For a fully drawn term loan it is close to the outstanding balance. For a revolving facility it is more: companies in trouble draw down their committed lines before they default, because the line is the last cash available. The EAD of a revolving facility is therefore the drawn amount plus a share of the undrawn amount, set by a credit conversion factor.
Expected loss combines the three
Multiply the three numbers and you obtain the expected loss, or EL: the average annual loss the bank expects on the facility, across many years and many similar loans.
EL = PD × LGD × EAD
Expected loss is a cost; unexpected loss needs capital
Expected loss is an average, and banks treat it as a cost of doing business: it is priced into the margin and, under accounting rules, provisioned. But actual losses vary around the average, and in a bad year they can be several times higher. The variation above the average is unexpected loss, and it is what the bank's capital exists to absorb.
The same three parameters therefore feed two different pots of money. Provisions, an accounting concept, are driven by expected loss. Regulatory capital, a prudential concept, is driven by unexpected loss. A change in a borrower's PD moves both.
A rating is a PD with a name
The master scale
Banks express their view of a borrower's creditworthiness as a rating grade, and every grade maps to a PD band on the bank's master scale. The rating is how humans talk about risk; the PD is how the models and the capital calculations use it.
| Internal grade | Indicative agency equivalent | Indicative one-year PD | Description |
|---|---|---|---|
| 1–2 | AAA to AA | Below 0.03% | Exceptionally strong |
| 3 | A | About 0.05% | Strong |
| 4 | BBB+ / BBB | About 0.15–0.25% | Good, investment grade |
| 5 | BBB− | About 0.40% | Adequate, lowest investment grade |
| 6 | BB | About 1% | Speculative, moderate risk |
| 7 | B | About 3–5% | Speculative, significant risk |
| 8 | CCC | 15% and above | Weak, watchlist |
| 9–10 | D | 100% | Default |
The figures are indicative. Each bank calibrates its own master scale from its own default experience, and regulators require the floor for any corporate PD used in capital calculations to be at least 0.05 per cent.
Internal, external and regulatory ratings
A company may carry several ratings at once. External ratings are published by agencies such as S&P Global, Moody's and Fitch for companies that issue bonds. Internal ratings are assigned by each bank to every borrower, rated or not, using its own models and analysts. In France, the Banque de France also assigns a cotation to many companies through its FIBEN database, used among other things to determine which loans are eligible as collateral for central-bank refinancing. A mid-cap that never issued a bond may therefore have three different assessments: none from the agencies, one from each of its banks, and one from the Banque de France.
Through-the-cycle versus point-in-time
Rating philosophies differ in how they treat the economic cycle. A through-the-cycle rating reflects the borrower's creditworthiness across a whole cycle and moves little when the economy turns. A point-in-time rating reflects the risk over the next year given current conditions, and moves with the economy. Regulatory capital models tend towards through-the-cycle, to avoid capital requirements swinging with the cycle. Accounting provisions under IFRS 9 require point-in-time, forward-looking estimates. Banks therefore often maintain two sets of PDs derived from the same rating.
The tension between the two philosophies has a name: procyclicality. If ratings and provisions worsen sharply in a downturn, banks must hold more capital and book more provisions exactly when their profits fall, which pushes them to lend less, which deepens the downturn. Through-the-cycle capital models, the countercyclical capital buffer and supervisory flexibility on provisions during the 2020 pandemic are all attempts to dampen that loop. You will meet it again in Part 5.
Risk-weighted assets turn the parameters into capital
Two approaches to capital
Under the European capital rules, a bank calculates the capital it needs for credit risk in one of two ways. Under the standardised approach, risk weights are fixed by regulation, according to the borrower's external rating where it has one; an unrated corporate typically attracts a 100 per cent weight, and certain investment-grade unrated corporates can receive less under the 2025 rules. Under the internal ratings-based approach, or IRB, the bank uses its own PD estimates — and, under the advanced version, its own LGD and EAD estimates — in a formula set by regulation.
Either way, the output is risk-weighted assets: the exposure multiplied by a risk weight. The bank must hold capital equal to a percentage of its risk-weighted assets, and the capital consumed by a loan is what its margin must pay a return on.
The output floor limits how low internal models can go
Since 2025 the final Basel III reforms have been phasing in an output floor in the European Union: a bank's total risk-weighted assets calculated with internal models may not fall below a set percentage — rising to 72.5 per cent once fully phased in — of what they would be under the standardised approach. For high-quality corporate lending, where internal models produce low risk weights, the floor can bind and raise capital. This puts pressure on precisely the thin-margin facilities, such as investment-grade revolving lines, that banks use to open relationships.
Large exposures cap concentration
Separately from capital, the European rules limit the bank's exposure to any single client or group of connected clients to 25 per cent of its Tier 1 capital. The limit applies to the risk group, not the legal entity, which is why the coverage module's discussion of risk hierarchies has a regulatory edge.
Provisions recognise expected loss in the accounts
IFRS 9 in one page
Under the IFRS 9 accounting standard, applicable since 1 January 2018, a bank must set aside a provision — an expected credit loss allowance — for every loan from the day it is booked. The size depends on which of three stages the loan is in.
A loan in Stage 1 is performing and its credit risk has not increased significantly since origination; the bank provisions for the losses expected from defaults in the next twelve months. A loan in Stage 2 has suffered a significant increase in credit risk since origination but is not in default; the bank provisions for losses expected over the loan's entire remaining life. A loan in Stage 3 is credit-impaired, broadly meaning in default; the bank provisions lifetime losses and recognises interest only on the net, provisioned amount.
The jump from Stage 1 to Stage 2 is the most consequential event in monitoring, because the provision can multiply several times overnight without the borrower having missed a payment. Part 2 works through it with Helion.
The margin must pay for four things
What a loan's price has to cover
A loan's all-in price — margin and fees — must cover four costs before the bank earns anything. It must cover the cost of funding: the bank borrows the money it lends, and its Treasury charges each business a funds transfer price that reflects the tenor and liquidity of the loan. It must cover the expected loss. It must cover operating costs: origination, servicing and monitoring. And it must pay a return on the capital consumed. The coverage module introduced RAROC as the test that combines all four; the structuring stage in Part 2 shows how the deal team adjusts the levers to pass it.
Lending is one of the most supervised activities in a bank
The rules that frame the lifecycle
| Framework | What it governs in the lifecycle |
|---|---|
| EBA Guidelines on loan origination and monitoring (EBA/GL/2020/06) | Credit-granting policies, creditworthiness assessment, pricing, collateral valuation, monitoring; applied to new loans from 30 June 2021 |
| Capital Requirements Regulation (CRR), as amended by CRR3 | Risk-weighted assets, IRB models, credit conversion factors, large exposures, the definition of default |
| IFRS 9 | Classification, staging and expected-credit-loss provisioning |
| EBA definition-of-default and forbearance standards | When a loan is in default, non-performing or forborne |
| ECB prudential provisioning expectations and the CRR NPL backstop | Minimum coverage of non-performing exposures as they age |
| AnaCredit (ECB Regulation 2016/867) | Loan-by-loan reporting of credit to legal entities in the euro area |
| National insolvency law (in France, the Commercial Code) | Pre-insolvency and insolvency procedures that shape the workout stage |
The EBA guidelines on loan origination and monitoring deserve special mention, because they turned many good practices into supervisory expectations: documented credit policies, an assessment of the borrower's ability to repay based on projections rather than history alone, consideration of ESG factors, independent collateral valuation, and ongoing monitoring with early-warning indicators. Much of Part 2 is, in effect, those guidelines in operational form.
Work these through before reading on. Full solutions are in the Answer Key.
1.1 Two ways out
For each loan, identify the first and the second way out: a five-year term loan to a profitable software company, unsecured; a loan to a property company secured on an office building let to a single tenant; a revolving facility to a retailer guaranteed by its parent.
Software company: the first way out is its operating cash flow; the second is weak, limited to its ranking as a senior unsecured creditor and whatever value the business has in insolvency. Property company: the first way out is the rent from the single tenant; the second is the building, which the bank can sell, although a building let to one tenant loses value if that tenant leaves. Retailer: the first way out is the retailer's cash flow; the second is the parent's guarantee, worth as much as the parent's own creditworthiness.
1.2 Expected loss
A 200 million euro revolving facility is 50 million drawn. The borrower's PD is 1.0 per cent, the LGD 45 per cent and the credit conversion factor on undrawn amounts 40 per cent. Compute the EAD and the expected loss.
EAD = 50 + 40% × 150 = 110 million euros. EL = 1.0% × 45% × 110 = 0.495 million euros a year.
1.3 The cost of capital
A 100 million euro term loan has a risk weight of 70 per cent. The bank holds 12 per cent capital against risk-weighted assets and requires 12 per cent after-tax return on it. What annual after-tax profit must the loan generate? If expected loss is 0.3 million, operating cost 0.2 million and tax 25 per cent, what annual revenue above the cost of funding does that imply?
Risk-weighted assets are 70 million, capital 12% × 70 = 8.4 million, required after-tax profit 12% × 8.4 = 1.008 million. Pre-tax, 1.008 / 0.75 = 1.344 million. Revenue above the cost of funding must be 1.344 + 0.3 + 0.2 = 1.844 million, a margin of about 1.84 per cent on 100 million.
1.4 Stages
A borrower rated grade 4 at origination is downgraded to grade 6, still paying on time. Another is 95 days late on a material payment. A third is unchanged. Place each in an IFRS 9 stage and say what horizon its provision covers.
The downgraded borrower, still paying, is in Stage 2 if the downgrade meets the bank's significant-increase test, as a three-notch move typically would; its provision covers lifetime expected losses. The borrower 95 days past due on a material amount is in default and in Stage 3; lifetime losses, with interest recognised on the net amount. The unchanged borrower stays in Stage 1, with twelve-month expected losses.
Explain to a friend who lends money to relatives why a bank cares about three separate numbers — how likely the borrower is to fail, how much it would lose if they did, and how much they would owe at that moment — rather than just asking "are they good for it?"
What are the first and second ways out of a loan, and which parameter measures each?
The first way out is the borrower's cash flow, measured by PD; the second is what the bank can recover, measured by LGD.
Name the eight stages of the lifecycle and the three phases they fall into.
Deciding: origination, credit analysis and rating, structuring, approval. Putting in place: execution and booking. Living with it: servicing, monitoring, workout and exit.
Why is the EAD of a revolving facility larger than its drawn balance?
Because borrowers in difficulty draw their committed lines before defaulting, so part of the undrawn amount is expected to be drawn by the time of default.
What is the difference between expected and unexpected loss, and which pot of money covers each?
Expected loss is the average loss, covered by pricing and provisions; unexpected loss is the variation above it, covered by capital.
What does the output floor do, and why does it matter for investment-grade lending?
It caps how far internal models can reduce risk-weighted assets below the standardised approach, at 72.5 per cent once phased in; it binds most on low-risk lending, where internal models produce low weights.
What happens to the provision when a loan moves from Stage 1 to Stage 2?
The provision horizon changes from twelve months to the remaining lifetime, so the provision usually rises sharply.
1
What is a loan's “second way out”?
Right. The first way out is the borrower's cash flow, measured by PD.
That is the first way out.
Distribution is a tool, not the second way out.
No.
Why is the EAD of a revolving facility larger than its drawn balance?
Right. Part of the undrawn amount is expected to be drawn by default.
It is.
Fees are not the reason.
No.
Which phase contains servicing, monitoring and workout?
Execution and booking.
Origination, analysis, structuring, approval.
They are its last phase.
Right. Deciding covers the first four stages; putting in place is execution and booking.
2The End-to-End Flow
This part follows Helion's financing package through all eight stages. The first four — origination, credit analysis, structuring and approval — decide whether and how the bank lends. The last four — booking, servicing, monitoring and workout — decide what the decision is worth.
Hold the whole flow in your head before the detail
One package, eight stages, about three years of story
The story runs as follows. In January, Helion's CFO tells the bank's RM that Helion intends to buy a US aerostructures company for about 400 million dollars and needs 250 million euros of bank financing alongside a bond. Over six weeks the bank analyses Helion, rates it, structures a term loan and a revolving facility, and takes the package to credit committee, which approves it with conditions. Over the following month the facility agreement is negotiated and signed, conditions precedent are satisfied, the facilities are booked and the term loan is drawn on the day the acquisition completes.
For a year and a half the loan is serviced uneventfully: interest periods roll, fees are billed, the first amortisation instalments are paid, and quarterly compliance certificates show comfortable covenant headroom. Then a supply-chain shock in the aerospace sector delays deliveries, Helion's earnings fall, leverage rises towards the covenant, early-warning signals fire, the facility moves to Stage 2 under IFRS 9 and onto the watchlist. The bank negotiates a covenant reset with a fee and a margin step-up. A year later Helion recovers, the loan returns to Stage 1, and eventually the package is refinanced and exits.
| Stage | Helion timeline | Key output |
|---|---|---|
| 1. Origination | January, week 1 | Opportunity logged, deal team formed |
| 2. Credit analysis and rating | Weeks 1–4 | Spread accounts, projections, grade 5, PD 0.40% |
| 3. Structuring | Weeks 2–5 | Term sheet: 150m amortising term loan, 100m RCF, covenants, pricing |
| 4. Approval | Week 6 | Conditional sanction by the regional credit committee |
| 5. Execution and booking | Weeks 7–10 | Signed agreement, CPs satisfied, facilities booked, first drawdown |
| 6. Servicing | Months 3 onwards | Interest periods, fees, amortisation, notices |
| 7. Monitoring | Continuous, quarterly tests | Covenant headroom, early-warning signals, annual review, staging |
| 8. Workout and exit | Month 20 onwards | Watchlist, covenant reset, cure, eventual refinancing |
Stage 1 — Origination brings the opportunity in
Coverage owns the first stage
The coverage module described origination in depth. For the lifecycle, three outputs of origination matter. The opportunity must be logged with its product, size and expected timing, so that the pipeline, the deal team and, later, the credit workflow all refer to the same deal. The client must be onboarded, or its KYC confirmed as current, before any commitment is made. And the RM must give credit an early indication of the request, so that analysis can start in parallel with structuring rather than after it.
At Helion, the RM logs a 250 million euro acquisition-financing opportunity, confirms that the KYC file is current, informs the credit analyst covering industrials, and forms a deal team with the loan structuring desk and the rates desk, which will propose hedging for the floating-rate debt.
Stage 2 — Credit analysis turns accounts into a rating
The question and the method
Credit analysis asks one question: can and will this borrower repay this loan? The method has five steps. The analyst gathers data. The analyst spreads the financial statements into a standard format. The analyst assesses business risk, the qualitative half, and financial risk, the quantitative half. The analyst assigns a rating. And the analyst writes a credit proposal — the deliverable on which the approval stage decides.
Step 1 — Data gathering
The inputs come from several sources of differing reliability. Audited annual financial statements are the backbone, with notes that matter as much as the main statements. Interim management accounts and the borrower's budget and business plan provide recency and forward view. Public data adds market and sector context: bond prices, agency ratings, filings, news. Internal data adds the bank's own experience: account conduct, payment behaviour, existing exposures. For acquisitions, the target's accounts, the purchase agreement and the due-diligence reports are added.
Step 2 — Spreading
Spreading means re-casting the borrower's accounts into the bank's standard template, so that every company is analysed on the same definitions. A spread separates recurring from non-recurring items, standardises the treatment of leases, pensions and factoring, and maps hundreds of accounting lines into a few dozen analytical ones. It is the most manual step in credit analysis and, for that reason, the most automatable; Part 4 returns to it.
Step 3a — Business risk: is this a good business?
Before the ratios, the analyst asks qualitative questions. How attractive and how cyclical is the industry? How strong is the company's competitive position: market share, technology, customer concentration, pricing power? How good and how credible is management, and how sound is governance? How diversified are its products, customers and geographies? What do ownership and the shareholders' financial policy imply for future leverage?
For Helion, the analyst notes long-term contracts with major aircraft manufacturers, good technology and a diversified energy business, but high customer concentration — the top three customers are 45 per cent of revenue — and exposure to the aerospace production cycle. The family shareholders have historically kept leverage moderate; the private-equity minority since 2023 may push for more.
Step 3b — Financial risk: leverage, coverage, cash flow and liquidity
The financial assessment turns the spread into ratios that each answer a question.
| Ratio | Formula | The question it asks | Helion |
|---|---|---|---|
| Net leverage | Net debt / EBITDA | How many years of earnings would repay the debt? | 900 / 420 = 2.1x |
| Interest cover | EBITDA / net interest | How comfortably do earnings cover interest? | 420 / 45 = 9.3x |
| FFO to debt | FFO / gross debt | What share of debt could cash from operations repay each year? | 305 / 1,150 = 27% |
| Gearing | Net debt / equity | How much of the business is funded by debt rather than owners? | 900 / 1,600 = 0.56 |
| Free cash flow | FFO − working capital − capex | Is there cash left after running and maintaining the business? | 135 positive |
| Liquidity | Cash + undrawn committed lines vs 12-month needs | Can it meet obligations for the next year even if markets close? | Cash 250 + undrawn lines; comfortable |
The analyst then builds projections: a base case and a downside case for the next five years, including the acquisition. In the base case, the acquisition adds about 50 million euros of EBITDA and about 400 million euros of net debt, so pro-forma leverage rises to 1,300 / 470 = 2.8x and falls back towards 2.2x within three years as the term loan amortises. In the downside case, a 15 per cent fall in aerospace volumes takes EBITDA to about 400 and leverage to about 3.3x.
Step 4 — Assigning the rating
The rating is assigned in three moves. A model — a scorecard calibrated on historical defaults for the relevant segment — produces a score from the financial ratios and the qualitative factors. The analyst may apply overrides where the model misses something material, within limits and with documented justification; banks monitor the override rate, because frequent overrides suggest either a weak model or analysts gaming it. Finally, notching adjusts for group support or country risk: a subsidiary may be rated up towards its parent if the parent is likely to support it, or capped by the rating of the country in which it operates.
At Helion, the model produces grade 5 on the base case. The analyst considers a one-notch downward override for customer concentration and acquisition risk, but concludes that the downside case still leaves leverage below 3.5x and that the family's track record supports the model grade. The final rating is grade 5, PD 0.40 per cent.
Obligor rating and facility rating
The rating just assigned is an obligor rating: it describes Helion and drives the PD. Many banks also assign a facility rating, or LGD grade, to each loan, reflecting its security, ranking and structure. The term loan, senior and guaranteed by the main operating subsidiaries, gets an LGD of 40 per cent; an unsecured loan to the holding company alone, structurally subordinated to creditors of the operating companies, might get 55 per cent. Two loans to the same borrower can therefore carry the same PD and different expected losses.
Step 5 — The credit proposal
Everything converges into one document: the credit proposal, also called the credit application or credit memo. It presents the request; the borrower, its business and its owners; the financial analysis and projections; the rating; the proposed structure, pricing and security; the risks and mitigants; the RAROC; compliance with policy and limits; and a recommendation. It is the document the approval stage decides on, and in most banks it is the single most important artefact in the bank's credit file — the one that supervisors and auditors read first.
A rating is only as good as the data behind it. Spreading feeds the rating engine; the engine applies the model; the rating and PD flow to the limit system, the pricing tool, the capital calculation and, at the next reporting date, the provisioning engine. Every step must be traceable: which financial statements, which version of the model, which overrides by whom and why. A rating whose inputs cannot be reproduced is, for a supervisor, not a rating at all.
Stage 3 — Structuring turns a yes into a shape
Structuring designs the loan around the risk
Credit analysis says how risky the borrower is. Structuring decides the shape of the loan so that it fits three things at once: the borrower's need, the bank's risk appetite and, for larger deals, what other lenders will accept. In practice analysis and structuring run in parallel and iterate. The structurer has four levers.
Lever 1 — Amount, tenor and amortisation
How much, for how long, and repaid how? The amount must match a genuine need and the borrower's capacity: a loan larger than the company can service creates risk without value. The tenor must match the use: working-capital lines are short, acquisition and capex loans longer, and longer tenors carry more risk because more can go wrong. The repayment profile can be amortising, with regular instalments that reduce exposure over time; bullet, with everything repaid at maturity and therefore refinancing risk at the end; or balloon, a mix in which part amortises and a large final payment remains.
Lever 2 — Pricing
The bank is paid in several ways. The margin is the spread over a reference rate — EURIBOR or €STR for euro loans, SOFR for dollars, SONIA for sterling — paid on drawn amounts. A commitment fee is paid on the undrawn part of a committed facility, typically expressed as a percentage of the margin, often 35 per cent. An upfront or arrangement fee is paid at signing. A utilisation fee may be added when a revolving facility is heavily drawn. A margin ratchet links the margin to a ratio, usually leverage, so that the price falls when the borrower improves and rises when it deteriorates.
Lever 3 — Security, guarantees and ranking
These are the second way out. Security gives the lender a claim over specific assets — shares, receivables, property, bank accounts. Guarantees add another obligor: typically the parent guarantees a subsidiary's loan, or operating subsidiaries guarantee a loan to the holding company, which addresses structural subordination, the fact that creditors of a holding company rank behind creditors of its subsidiaries for the subsidiaries' assets. Ranking decides the order of repayment in insolvency: senior before subordinated, secured before unsecured. Investment-grade corporate loans are usually unsecured but supported by guarantees and a negative pledge — a promise not to grant security to other creditors.
Lever 4 — Covenants and undertakings
Covenants are the loan's rules and its early-warning system. Financial covenants are ratios tested at regular intervals: maximum net leverage, minimum interest cover, sometimes minimum liquidity or a cap on capital expenditure. Information undertakings require the borrower to provide accounts, compliance certificates and notice of material events. General undertakings restrict behaviour: no disposals of key assets, no additional security for other lenders, no change of business, limits on additional debt. Events of default list the circumstances — non-payment, covenant breach, insolvency, cross-default on other debt, change of control in some loans — that allow lenders to accelerate.
Covenants matter not because the bank wants to call the loan at the first breach — it almost never does — but because a breach gives the bank a seat at the table: the right to renegotiate before the problem becomes a default.
Covenant packages differ sharply between markets. Investment-grade loans typically carry one or two financial covenants, or none for the strongest borrowers. Leveraged loans to private-equity-owned companies historically carried several maintenance covenants, but the institutional leveraged-loan market has moved largely to covenant-lite structures with no maintenance covenant at all, only incurrence tests that apply when the borrower takes a specific action such as raising more debt. The trade-off is clear: fewer covenants mean fewer early seats at the table, and lower recoveries when things go wrong. The Corporate and Leveraged Finance module develops this.
Bilateral, club or syndicated is also a structuring choice
Helion's package is small enough for the bank to hold alone, and the bank wants the relationship credit of a bilateral deal. Had Helion asked for 1 billion euros, the bank could not have held it all: its single-name appetite and the large-exposure rules would cap its hold, and the structure would have become a club or a syndicated loan, with the bank as arranger and a share sold to other lenders. Syndication adds an underwriting decision — whether the arranger commits to the full amount and bears the risk of not selling it — and a documentation standard, usually the Loan Market Association templates in Europe. The Distribution module develops the syndication process.
Stage 4 — Approval decides who says yes
Credit approval is a governed decision, not an opinion
Credit approval is the controlled process by which a bank decides, and documents, whether to commit its balance sheet. Its foundations are set by the board through the risk appetite framework: how much credit risk the bank wants, in which sectors, countries and rating bands, and with what concentration. Approval enforces that appetite deal by deal.
Delegation of authority routes each deal to the right level
Not every deal goes to the top. Authority is delegated according to risk, size and complexity: a small, well-rated loan can be approved by a single senior credit officer, a large or risky one needs a regional committee, and the largest exposures go to the group credit committee or even the board's risk committee.
The delegation matrix has several dimensions at once: the total exposure to the group, the rating, the tenor, the product, whether policy exceptions are requested, whether the borrower is on the watchlist, and sometimes sector or country. The strictest dimension decides the level: a small loan that requires a policy exception goes to a higher authority than its size alone would suggest.
| Total group exposure | Grades 1–4 | Grade 5 | Grades 6–7 | Watchlist or grade 8 |
|---|---|---|---|---|
| Up to 25m | Senior credit officer | Senior credit officer | Head of regional credit | Regional committee |
| 25m to 150m | Head of regional credit | Regional committee | Regional committee | Group committee |
| 150m to 500m | Regional committee | Group committee | Group committee | Group committee |
| Above 500m | Group committee | Group committee | Board risk committee | Board risk committee |
Helion's total exposure after the new package will be the 250 million of new facilities plus 120 million of existing hedging and guarantee lines, so 370 million euros, at grade 5. The matrix routes it to the group committee. Illustrative matrix.
The four-eyes principle and the independence of sanction
The single most important structural control in approval is that whoever profits from the deal cannot be the one who approves it. The RM proposes; an independent credit officer, in the second line of defence, recommends; an authorised person or committee sanctions. This four-eyes principle runs through the whole lifecycle: the person who books a loan is not the person who checks it, and the person who releases funds is not the person who requested them.
The credit committee is a structured, minuted decision
A credit committee is not an informal discussion. It has terms of reference, a quorum, voting rules — often requiring the head of risk's agreement, or giving risk a veto — and minutes that record the decision and the reasons. Its value lies not in approving deals but in the quality of its questions: what happens to repayment in the downside case, why this structure rather than a tighter one, what the bank knows about the acquired company's customers, whether the pricing pays for the risk.
Three outcomes, and most decisions are conditional
A committee can approve, decline or — most often — approve with conditions. Conditions come in two families. Conditions precedent must be satisfied before money moves: signed guarantees, legal opinions, evidence that the acquisition has completed, satisfactory KYC on new entities. Conditions subsequent and covenants must remain satisfied throughout the loan's life: delivery of accounts within 120 days, financial ratios, a hedging strategy within three months. A committee may also tighten the structure: reduce the amount, shorten the tenor, add a covenant or require security.
The sanction becomes the source of truth for everything downstream
Once decided, the sanction is recorded in the credit workflow system as a formal record: the approved limit, structure, pricing range, conditions, covenants and expiry. That record drives the legal documentation, the limit set-up and the covenant capture in the next stage. If what is signed differs from what was sanctioned — a covenant loosened in negotiation, a guarantor dropped — the difference must go back for approval. Gaps between the sanction and the signed agreement are one of the most frequent findings in credit audits.
Modern banks run approval on a loan origination system — a credit workflow platform such as nCino, Moody's CreditLens or an in-house tool — that holds the proposal, routes it through the delegation engine, records each recommendation and decision with a time-stamp, and produces the sanction record. The delegation engine is the paper matrix turned into executable rules: it reads the deal's attributes and routes the proposal to the right authority, and it must be able to prove, years later, that the decision was taken by someone entitled to take it. That audit trail is what supervisors test.
Stage 5 — Execution and booking make the decision real
From sanction to signed contract
The sanction is a decision; the facility agreement is the contract. Lawyers — the bank's internal legal team and usually external counsel for both sides — turn the term sheet into a full agreement. In Europe, most corporate facility agreements follow the templates of the Loan Market Association (LMA), which provide standard clauses for representations, undertakings, events of default, interest mechanics and agency, leaving the commercial terms to be negotiated. Using a market standard reduces negotiation time and, crucially for operations, means that every agreement in the bank's book has the same structure.
Negotiation changes things. The borrower pushes for wider covenant headroom, fewer restrictions, more flexibility. Every change from the sanction must be checked against what was approved, and anything outside the sanction must be re-approved.
Conditions precedent are the last gate before money moves
Signing does not release funds. First, every condition precedent must be satisfied and evidenced: corporate authorisations, signed guarantees, legal opinions on capacity and enforceability, KYC on all obligors and guarantors, evidence of the acquisition's completion, the agreed hedging, payment of fees. A CP checklist is maintained, each item ticked with evidence by legal or loan operations and checked by a second person. Only when every CP is satisfied, or formally waived by an authorised person, are funds released.
Booking turns the contract into data
Booking is the act that makes the signed deal real on the bank's books: the legal contract becomes a live asset in the loan servicing system. It is owned by middle and back office, not by the front office, precisely to keep the people who negotiate a deal separate from the people who record it.
Every loan system is built on a three-level hierarchy. The facility is the legal commitment: Facility B, a 100 million euro RCF to Helion maturing in five years. Under it sit one or more tranches where a facility has distinct sub-limits, currencies or borrowers. Under each tranche sit the drawings, also called outstandings, loans or contracts: each individual advance, with its own amount, currency, interest period and rate. A revolving facility might have dozens of drawings over its life; a term loan usually one or a few.
Six booking steps
| Step | What happens | Main control |
|---|---|---|
| 1. Deal capture | Terms from the signed agreement are entered as structured fields | Four-eyes check against the agreement and the sanction |
| 2. Facility and tranche set-up | Amounts, currencies, dates, purpose, borrowers and guarantors created | Reconciliation to the term sheet |
| 3. Static data | Counterparty, product, calendars, day-count, rate index and payment instructions linked | Golden-source references only, no free typing |
| 4. Limit activation | The facility is linked to the approved credit limit so that drawings consume it | Limit amount cannot exceed the sanction |
| 5. CP check | All conditions precedent confirmed and evidenced | Second-person sign-off; no drawdown before release |
| 6. First drawdown | Drawdown request validated, rate fixed, funds paid, position recorded | Payment release under dual authorisation |
Booking feeds six downstream consumers
The booked record is the single upstream source for most of the bank. Servicing reads it to calculate interest and fees. Payments read it to move money. Credit risk reads it to measure utilisation against limits and exposure. Finance reads it to account for the loan at amortised cost, spread fees over its life under the effective-interest method, and calculate provisions. Regulatory reporting reads it for capital, large exposures and loan-level returns such as AnaCredit. And covenant monitoring reads it to know which tests apply when. An error at booking is not a booking error; it is an error in all six places at once.
Corporate loans remain one of the most manual areas of banking. Facility agreements arrive as long PDF documents; operations staff read them and key the terms into the loan system, often again into the limit system, and again into a covenant tracker. Each re-keying is a chance for error, and a reconciliation is needed afterwards to find it. Loan servicing platforms such as Finastra's Loan IQ and FIS's ACBS dominate large corporate books; the frontier is extracting terms from the document as structured data, with lineage back to the clause, so that the same data feeds every downstream system. Part 4 opens the box.
Stage 6 — Servicing runs the loan for years
Servicing is the longest phase
Origination is a moment; servicing is the marathon. From first drawdown to final repayment, the loan must be run correctly every day: interest calculated and collected, fees billed, drawdowns processed, repayments received, rates reset, notices sent, amendments made. For a five-year revolving facility, that can mean hundreds of events, each of which changes the loan's state and must be right.
Every euro is principal, interest or a fee
Every cash flow on a loan is one of three things. Principal moves the outstanding balance: drawdowns increase it, repayments and prepayments reduce it. Interest is the price of the money drawn. Fees are the price of everything else: the commitment, the arrangement, the agency, the waiver.
Interest is rate times balance times time
Interest on a drawing is the product of four things: the rate, the balance, the time elapsed and a day-count convention that defines how days and years are counted.
Interest = principal × (reference rate + margin) × days / year basis
For floating-rate loans, the rate is reset at the start of each interest period, typically one, three or six months, chosen by the borrower. With a term rate such as EURIBOR, the rate is fixed in advance, usually two business days before the period starts, and the interest for the period is known from day one. With an overnight risk-free rate such as €STR, SOFR or SONIA, the rate is compounded day by day in arrears and the interest is known only at the end of the period, so market conventions add a short lookback — typically five business days — to give operations time to calculate and notify the payment.
Drawdowns, rollovers and repayments
Under a revolving facility, the borrower sends a utilisation request a few business days before it wants the money, specifying amount, currency and interest period. Operations checks that the request is valid: within the available commitment, no event of default outstanding, conditions met. Then it fixes the rate, books the drawing and pays the funds. At the end of each period a drawing is either repaid or rolled over into a new period at a new rate; a rollover in which an old drawing is replaced by a new one is often settled net, so that only the interest and any difference in principal move.
Term loans are repaid according to their schedule. Helion's Facility A amortises 15 million euros a year, so each anniversary operations bills the instalment and reduces the outstanding. Voluntary prepayments are allowed on most corporate loans, usually only at the end of an interest period or with break costs that compensate the lender for funding it has to unwind. Mandatory prepayments are triggered by events defined in the agreement — the proceeds of a major disposal, a change of control, illegality.
Fees run on their own calendars
| Fee | Basis | When it is paid |
|---|---|---|
| Arrangement or upfront fee | Percentage of the facility amount | At signing or first drawdown |
| Commitment fee | Percentage of the margin, on undrawn committed amounts | Quarterly in arrears |
| Utilisation fee | Additional margin when drawings exceed set thresholds | With interest |
| Agency fee | Fixed annual amount, syndicated loans only | Annually in advance |
| Amendment or waiver fee | Negotiated percentage or fixed amount | On effectiveness of the amendment |
| Prepayment fee or break costs | Contractual percentage or actual funding loss | On prepayment |
Notices, payments and reconciliation
Every servicing event generates a formal, dated notice to the borrower: rate-fixing notices, interest-due notices, fee invoices, repayment reminders. Notices are both a contractual obligation and an audit trail. The payments themselves move through the bank's payment systems — in the euro area via TARGET2 or the T2 system for large-value payments and SEPA for smaller ones, with SWIFT messages moving between banks — and every expected cash flow must be reconciled against what actually arrived. Unmatched items, called breaks, must be investigated: a borrower that paid the wrong amount, a missed rate reset, a payment to the wrong account.
Amendments and lifecycle events
Over a loan's life, the agreement itself changes. The borrower asks for an increase, an extension of maturity, a change of covenant, the addition of a new subsidiary as borrower, a waiver of a breach. Each amendment is a controlled transaction: credit approval if the risk changes, legal documentation, booking in the system with an effective date, and recalculation of everything downstream.
In a syndicated loan, most servicing is performed by the agent bank on behalf of all lenders. The agent receives drawdown requests and notifies lenders, collects each lender's share of the funds, fixes rates, calculates interest and fees, collects payments from the borrower and distributes them to lenders pro rata. Each lender still books and services its own share in its own system, and must reconcile its records with the agent's notices. The Agency module develops this; the key point here is that in a syndicated loan the same event is processed twice, once by the agent and once by each lender, and the two must agree.
Stage 7 — Monitoring keeps the rating and the provisions honest
Monitoring is continuous, not annual
Credit analysis at origination is a snapshot. Monitoring is the continuous process of checking whether the snapshot is still true. Its outputs are a rating that reflects current risk, provisions that reflect current expected loss, and early action when things deteriorate. It combines four streams of information: financial information — accounts and compliance certificates; behavioural information — utilisation, payment patterns and account flows; market information — share and bond prices, agency ratings, sector news; and qualitative information — what the RM learns from the client.
Covenants are the first structured signal
Covenant monitoring runs a loop: capture each covenant and its definitions from the documentation; build a calendar of test dates; collect the borrower's compliance certificate — a signed statement, usually from the CFO, setting out the covenant calculations — and the supporting accounts; test the figures, ideally by recomputing them rather than relying on the borrower's own calculation; grade the result; and escalate when needed.
Grading matters more than pass or fail. A covenant met with 40 per cent headroom is a comfortable credit; one met with 3 per cent headroom is a credit on the edge, even though it technically passes. Many banks grade headroom: above 20 per cent green, between 10 and 20 per cent amber, below 10 per cent red, breach black.
Definitions decide the answer
Covenant ratios depend entirely on the definitions in the agreement. Is EBITDA adjusted for "non-recurring" items, and who decides what is non-recurring? Are leases included in debt? Is supplier finance — money owed to a bank that paid the company's suppliers early — counted as debt or as trade payables? Are the earnings of an acquired business included pro forma for the full year? Two analysts can compute leverage of 2.9x and 3.6x for the same company from the same accounts, both correctly, under two different definitions. This is why covenant capture is the hardest step in monitoring and why the definitions must be captured with the numbers.
Early-warning signals catch what covenants miss
Covenants are tested only every quarter or half-year, on figures that are already weeks old. Early-warning systems combine faster signals.
| Signal family | Examples | Why it is useful |
|---|---|---|
| Behavioural | Rising utilisation of revolving lines, late payments, overdrafts, reduced incoming flows | Daily data, often the first sign of liquidity stress |
| Financial | Falling margins, rising leverage, negative free cash flow, auditor qualifications | Structural deterioration |
| Market | Share-price falls, widening bond or CDS spreads, agency downgrades or negative outlooks | The market's real-time view |
| Qualitative | Management departures, loss of a major customer, litigation, delayed accounts | Context the numbers do not show |
| Sector and macro | Sector downturns, commodity shocks, sanctions | Portfolio-wide risk |
At Helion, early-warning signals fire before the month-18 certificate arrives: revolving-facility utilisation rises from 20 to 60 million euros in two months, Helion's bond spread widens by 120 basis points after a profit warning, and a news alert reports a production halt at its largest customer.
Re-rating: the calendar and the world
Two clocks keep a rating current. The periodic clock requires every borrower to be re-rated at least annually, at the annual review, when the credit file, limits and conditions are refreshed. The event-driven clock requires an immediate re-rating when something material happens: a covenant near-miss, a profit warning, a downgrade, an acquisition. After the month-18 test, the analyst re-rates Helion from grade 5 to grade 7, with a one-year PD of about 3 per cent.
IFRS 9 staging: the jump from twelve months to lifetime
The downgrade triggers the most consequential accounting event in monitoring. Under IFRS 9, a loan moves from Stage 1 to Stage 2 when its credit risk has increased significantly since origination. Banks define that SICR test in their policies, typically through a combination of a relative increase in lifetime PD — for example, a doubling or tripling since origination — an absolute increase, qualitative triggers such as watchlist status or forbearance, and a backstop at 30 days past due.
When a loan becomes non-performing
At the next step down the ladder, a loan is in default under the European definition when the borrower is more than 90 days past due on a material amount — for corporates, more than 500 euros and more than 1 per cent of its exposure — or when the bank judges the borrower unlikely to pay in full without recourse to collateral. Indicators of unlikeliness to pay include a distressed restructuring, the bank's own sale of the loan at a large discount, a specific provision, or an insolvency filing. A defaulted loan is non-performing and in IFRS 9 Stage 3.
Forbearance is a concession to a borrower in difficulty
Forbearance means granting a concession — an extension, a covenant waiver, a payment holiday, a rate reduction — to a borrower that is in financial difficulty, in terms the bank would not grant to a healthy borrower. Forbearance must be flagged in the bank's systems and reported, because it can mask deterioration. A forborne exposure that remains performing stays flagged for a probation period of at least two years; a forborne non-performing exposure must show at least one year of good behaviour before it can return to performing. The flag, not the loan's current behaviour, drives much of the regulatory reporting.
Portfolio monitoring watches the forest as well as the trees
Single-name monitoring watches each borrower. Portfolio monitoring watches the shape of the book: concentrations by sector, country and rating; rating migration, the share of borrowers moving up or down each year; cohort analysis, which follows loans originated in the same period; and stress testing, which asks what would happen to ratings, provisions and capital under a severe but plausible scenario. In a sector shock like Helion's, portfolio monitoring is what tells the bank that Helion is not alone: every aerospace supplier in the book is deteriorating at once.
Monitoring is a data pipeline stitching the whole lifecycle together: booked exposures from the loan system, ratings from the rating engine, covenant results from the tracker, behavioural data from payments and accounts, market data from vendors, and macroeconomic scenarios from economists, all feeding an ECL engine that stages and provisions every exposure at every reporting date. The output must reconcile to the general ledger and to regulatory returns. The Basel principles for risk data aggregation, BCBS 239, exist because many banks discovered in 2008 that they could not add up their own exposures quickly and correctly. They are still among the hardest supervisory expectations to meet.
Stage 8 — Workout and exit: how every loan ends
Most loans exit quietly
Most loans end the way they were meant to. Term loans are repaid on schedule; revolving facilities expire or, more often, are refinanced, usually by the same banks on new terms a year or two before maturity; acquisition loans are taken out by bonds or equity. A clean exit still has work: final interest and fees, cancellation of the limit, release of guarantees and security, closure in the loan and limit systems, archiving of the credit file. Facilities that are never formally closed leave phantom limits and phantom exposures in the data, which distort capital and concentration figures for years.
Some loans go onto the watchlist
When monitoring shows a significant deterioration, the borrower goes onto the watchlist: a list of names under heightened attention. Watchlist status triggers more frequent reviews, a higher approval level for any new exposure, an action plan agreed with Risk, and usually IFRS 9 Stage 2. Most watchlist names recover; a minority deteriorate further and are transferred to the workout unit, a specialist team — sometimes called special assets or restructuring — that takes over the relationship from coverage.
The mindset shift: from PD to recovery
Once default becomes likely, the question changes. At origination and in monitoring, the bank asked how likely the borrower was to fail. In workout, it asks how to maximise the recovery — the present value of what it will eventually get back. Workout is not about avoiding a loss at any cost; it is about choosing the path with the highest risk-adjusted, discounted recovery.
The central fork is between going-concern and gone-concern value. A business that is still viable is usually worth far more alive — continuing to trade, keeping customers and employees — than broken up and sold asset by asset. Most workouts therefore try to preserve the going concern.
Five strategic paths
| Path | What it involves | When it fits |
|---|---|---|
| Wait and monitor | Intensive monitoring, information rights, no change to terms | Temporary problem, strong sponsor, comfortable liquidity |
| Waiver and amend-and-extend | Waive a breach, reset covenants, extend maturities, often for a fee and a higher margin | Viable business, problem of timing rather than solvency |
| Deeper restructuring | New money, debt write-down, debt-for-equity swap, asset sales | Viable business with too much debt |
| Sale of the loan | Sell the exposure to a distressed-debt investor at a discount | The bank wants certainty or lacks the capacity to run the workout |
| Enforcement and insolvency | Accelerate, enforce security, or file under formal insolvency procedures | Cooperation has failed or the business is not viable |
Restructuring tools, from light to heavy
A standstill agreement buys time: lenders agree not to enforce for a period while a plan is negotiated. A waiver forgives a specific breach; an amendment changes the terms going forward. Amend-and-extend pushes maturities out, usually in exchange for higher pricing and tighter terms. New money — additional financing, often ranking ahead of existing debt — may be needed to keep the business running. A debt-for-equity swap converts part of the debt into shares, reducing leverage and giving lenders upside if the business recovers. A haircut writes down part of the debt. These tools are combined in a restructuring plan; the Debt Restructuring & Advisory module develops them, together with the multi-creditor dynamics that make large restructurings so difficult.
Formal procedures in France
When consensual restructuring fails, the courts take over. French law offers a graded set of procedures. Mandat ad hoc and conciliation are confidential, amicable pre-insolvency procedures in which a court-appointed person helps the company negotiate with its creditors. Sauvegarde, for a company in difficulty but not yet insolvent, and its accelerated variant, protect the company while a plan is adopted. Redressement judiciaire applies when the company is in cessation des paiements — unable to pay its debts as they fall due from its available assets — and liquidation judiciaire when recovery is impossible. Since a 2021 reform transposing the EU Restructuring Directive, larger cases group creditors into classes that vote on the plan, and a plan can in certain conditions be imposed on dissenting classes, a cross-class cram-down. For lenders, the procedure determines who controls the outcome and how security can be enforced.
Recovery, provisioning and write-off
Every workout has an accounting shadow. In Stage 3, the provision is assessed individually for larger exposures, usually as the difference between the exposure and the present value of the expected recoveries under several scenarios. As recoveries come in, they are recorded against the exposure. When there is no reasonable expectation of further recovery, the remaining balance is written off. Supervisory rules add a backstop: under the European NPL backstop for loans originated since April 2019, non-performing exposures must be fully covered by provisions or capital deductions after three years if unsecured, and after seven to nine years if secured, whatever the bank's own recovery estimate.
Three ways out of workout
Every workout ends one of three ways. The borrower cures: it recovers, the loan returns to performing after the required probation, and the relationship goes back to coverage, as Helion's does. The bank sells the debt, individually or as part of a portfolio of non-performing loans, accepting a discount for certainty and capital relief; the EU's 2021 directive on credit servicers and credit purchasers was designed to deepen that market. Or the case is closed after enforcement, insolvency and final recovery, with the remainder written off.
Work these through before reading on. Full solutions are in the Answer Key.
2.1 Spreading and ratios
A company reports EBITDA of 180 million euros after a one-off gain of 20 million on an asset sale, gross debt of 650 million including 50 million of leases, cash of 110 million, net interest of 30 million and tax of 25 million. Compute adjusted EBITDA, net debt, net leverage, interest cover and FFO.
Adjusted EBITDA = 180 − 20 = 160 million. Net debt = 650 − 110 = 540 million. Net leverage = 540 / 160 = 3.4x. Interest cover = 160 / 30 = 5.3x. FFO = 160 − 30 − 25 = 105 million.
2.2 Routing
Using the illustrative delegation matrix in this part, route three requests: a 20 million euro facility to a grade-4 borrower with no existing exposure; a 30 million increase for a grade-6 borrower with 100 million of existing exposure; a 10 million facility to a watchlist borrower.
The 20 million facility to a grade-4 borrower with no other exposure goes to a senior credit officer. The 30 million increase for a grade-6 borrower takes total exposure to 130 million, so it goes to the regional committee. The 10 million facility to a watchlist borrower goes to the regional committee, because the watchlist column is the strictest dimension even at a small amount.
2.3 Interest and fees
A 40 million euro drawing under a revolving facility runs for 92 days at three-month EURIBOR of 2.25 per cent plus a margin of 1.10 per cent, actual/360. The facility is 100 million euros and the commitment fee is 35 per cent of the margin on undrawn amounts. Compute the interest on the drawing and the commitment fee for the same 92 days.
Interest = 40,000,000 × (2.25% + 1.10%) × 92 / 360 = 40,000,000 × 3.35% × 92 / 360 = 342,444.44 euros. The undrawn amount is 60 million and the commitment fee rate is 35% × 1.10% = 0.385 per cent: 60,000,000 × 0.385% × 92 / 360 = 59,033.33 euros.
2.4 Headroom
A borrower's leverage covenant is 3.0x. Net debt is 540 million and rolling EBITDA 210 million. Compute leverage and EBITDA headroom, and grade it on the green-amber-red scale in this part.
Leverage = 540 / 210 = 2.57x. The EBITDA at which the covenant would be breached is 540 / 3.0 = 180 million, so headroom is 1 − 180 / 210 = 14 per cent: amber.
2.5 Staging
A loan of 50 million euros, LGD 45 per cent, has a twelve-month PD of 0.5 per cent at origination and a remaining-life PD of 2 per cent. After a downgrade, its twelve-month PD is 2.5 per cent and its remaining-life PD 8 per cent. Compute the provision before and after, assuming the downgrade triggers a move to Stage 2. By what multiple did it rise?
Before: Stage 1, twelve-month ECL = 0.5% × 45% × 50 = 0.1125 million euros. After: Stage 2, lifetime ECL = 8% × 45% × 50 = 1.8 million euros. The provision rose sixteenfold.
2.6 Workout choice
A 60 million euro exposure can be sold today at 70 per cent. Alternatively, a restructuring gives a 60 per cent chance of full recovery in present value of 57 million, and a 40 per cent chance of recovering 25 million. Which option maximises expected value, and what non-financial considerations might change the decision?
Sale: 70% × 60 = 42 million, certain. Restructuring: 0.6 × 57 + 0.4 × 25 = 34.2 + 10 = 44.2 million, uncertain. The restructuring has a slightly higher expected value. Considerations that could reverse the decision: the bank's capacity to run a long workout, the capital and provision relief of selling now, the uncertainty of the estimates, the relationship and reputational value of supporting a client, and whether other creditors' behaviour could lower the restructuring outcome.
Explain to a friend who has a mortgage why the bank might set aside much more money against their loan this quarter even though they have never missed a payment — and why that does not mean the bank thinks they will default.
What are the five steps of credit analysis, and what is the deliverable?
Data gathering, spreading, business and financial risk assessment, rating, and the credit proposal, which is the deliverable.
Name the four structuring levers.
Amount, tenor and amortisation; pricing; security, guarantees and ranking; covenants and undertakings.
What is the four-eyes principle, and where does it appear in the lifecycle?
The person who benefits from an outcome may not be the one who authorises it; it appears in approval, in booking checks, in CP sign-off and in payment release.
Distinguish conditions precedent from covenants.
Conditions precedent must be satisfied before funds are released; covenants must remain satisfied throughout the life of the loan.
Describe the facility–tranche–drawing hierarchy.
The facility is the legal commitment and holds the limit; tranches are distinct sub-limits within it; drawings are individual advances with their own amounts, periods and rates.
Why does a compounded risk-free rate need a lookback when a term rate does not?
Because the compounded rate is known only at the end of the period; the lookback shifts the observation period earlier so the amount is known a few days before payment.
What triggers a move from IFRS 9 Stage 1 to Stage 2, and what changes in the provision?
A significant increase in credit risk since origination, measured by the bank's SICR criteria with a 30-days-past-due backstop; the provision changes from twelve-month to lifetime expected loss.
What two tests define default under the European rules?
More than 90 days past due on a material amount, or the bank's judgement that the borrower is unlikely to pay in full without recourse to collateral.
What is the difference between going-concern and gone-concern value, and why does it matter in workout?
Going-concern value is what the business is worth if it keeps operating; gone-concern value is what its assets fetch if it is broken up. Going-concern value is usually much higher, so workouts try to preserve it.
2
What is the deliverable of credit analysis?
No.
That is documentation.
Right. After data gathering, spreading, risk assessment and rating.
That comes from structuring.
Which is NOT one of the four structuring levers?
A lever.
A lever.
A lever.
Right. Levers: amount/tenor/amortisation, pricing, security and ranking, covenants.
What is the four-eyes principle?
Right. It appears in approval, booking checks, CP sign-off and payment release.
No.
No.
No.
3Roles, Actors, and Interactions
A loan passes through more hands than almost any other product in the bank. This part names every player along the lifecycle, inside and outside the bank, explains why the bank deliberately separates them, and shows where their incentives pull in different directions.
The baton passes along the lifecycle
Each stage has an owner, and none owns them all
| Stage | Leads | Contributes | Independent check |
|---|---|---|---|
| Origination | Relationship manager | Product, structuring | KYC and financial security |
| Credit analysis and rating | Credit analyst | RM for client information | Rating review, model validation |
| Structuring | Structurer or product banker | RM, credit analyst, Treasury for funding | Credit officer |
| Approval | Credit officer and committees | RM, analyst, structurer | Risk veto, audit of the delegation |
| Execution and booking | Legal; loan operations (middle office) | RM, external counsel | Four-eyes on capture, CP sign-off |
| Servicing | Loan operations (back office); agent for syndicated loans | Payments, Treasury | Reconciliation, operational-risk controls |
| Monitoring | Credit risk monitoring; RM | Covenant team, finance for provisions | Provisioning committee, internal audit |
| Workout and exit | Workout unit; RM for normal exits | Legal, external advisers, finance | Workout and NPL committees |
The lifecycle is designed for separation
The table shows a deliberate pattern. At almost every stage, the person who wants the outcome is separated from the person who controls it. The RM who wants the deal does not rate the borrower or approve the loan. The structurer who negotiated the terms does not book them. The operations clerk who keys the drawdown does not release the payment. The workout specialist who estimates the recovery does not set the provision alone. Each separation exists because, at some point in banking history, its absence led to losses.
The front office originates and shapes
Relationship manager and structurer
The relationship manager owns the client and brings the opportunity, as described in the coverage module. In the lifecycle, the RM's specific responsibilities are to provide client information to credit, to manage the client's expectations during approval, to keep watch on the client throughout monitoring, and to be the first to raise concerns — the first line of defence.
The structurer or product banker designs the facility and its pricing, negotiates the term sheet with the client and, for syndicated deals, works with the syndicate desk on what the market will accept. Structurers are measured on deals closed and revenue; their natural bias is towards terms the client will accept.
Credit decides independently
Analysts, credit officers and committees
The credit analyst spreads the accounts, analyses the borrower, proposes a rating and writes or contributes to the credit proposal. In some banks analysts sit in the front office as first-line support to coverage; in others, and increasingly, they sit in the risk function to protect their independence. Either way, the rating they propose is reviewed by someone outside the deal team.
The credit officer — sometimes called the risk manager or sanctioning officer — sits in the second line of defence. The credit officer forms an independent view, recommends or sanctions within delegated authority, and sets conditions. The credit officer's natural bias is towards protection: tighter terms, lower amounts.
Credit committees take the larger decisions. Banks run several kinds.
| Committee | What it decides | Typical members |
|---|---|---|
| Credit committee (regional, group) | New facilities, increases, renewals above delegated limits | Chief risk officer or delegate, senior credit officers, business heads without a vote or with a non-decisive vote |
| Underwriting committee | Commitments to underwrite syndicated deals | Credit, syndicate, capital markets heads |
| Watchlist committee | Names to add, keep or remove; action plans | Credit risk, coverage, workout |
| Provisioning committee | Staging overrides, individual provisions, management overlays | Finance, credit risk, CFO delegate |
| Workout or NPL committee | Restructuring terms, sales, write-offs | Workout head, credit, legal, finance |
| Model risk committee | Approval and changes of rating and provisioning models | Model risk management, risk heads, validation |
Legal turns terms into enforceable promises
Internal counsel, external counsel and the LMA
Internal legal teams review documentation, manage external counsel and advise on enforceability. For significant facilities, the bank instructs external counsel, who draft and negotiate the facility agreement, and the borrower has its own. Legal opinions from counsel confirm that the borrower and guarantors have the capacity to enter the agreement and that it is enforceable under the governing law. The Loan Market Association, the trade body for the European syndicated loan market, publishes the recommended templates most corporate facilities are based on, and the equivalent bodies in the United States and Asia do the same for their markets.
Loan operations make the loan run
Middle office and back office
Loan operations is usually split in two. The middle office sits closest to the deal: it captures the terms, sets up the facility, links it to the limit, checks conditions precedent and manages amendments. The back office runs the recurring machinery: rate fixings, interest and fee calculations, drawdowns, repayments, payments, notices and reconciliations. The split matters for control: the team that sets up the deal is not the team that moves the money.
The agent bank
In syndicated loans, one bank acts as agent for all lenders, collecting from the borrower and distributing to lenders, fixing rates, circulating information and coordinating votes on waivers and amendments. Some banks run large agency businesses for third parties; others act as agent only on deals they arrange. Independent agency firms have grown alongside them. The Agency module develops the role.
Risk, finance and the workout unit take over as the loan ages
Monitoring is shared between the lines
In monitoring, the first line — the RM and the business — watches the client, reports concerns and owns the action plan. The second line — credit risk monitoring — owns the rating, the watchlist decision and the early-warning framework, and challenges the first line when it is too optimistic. Finance owns the provisions: it runs the ECL engine, books the charge and defends the numbers to auditors. The provisioning committee brings the three together, because the ECL figure is part model and part judgement, and judgement is exactly where optimism and pessimism compete.
The workout specialist is a different animal
Workout calls for skills almost opposite to origination. An originator builds relationships, finds reasons to lend and wins mandates. A workout specialist assumes the relationship has failed, reads documents for every right the bank has, values assets for what they would fetch in a forced sale, negotiates with other creditors who are both allies and competitors, and is prepared to enforce. This is why most banks transfer distressed names out of coverage into a separate unit with its own reporting line.
| Dimension | Originator | Workout specialist |
|---|---|---|
| Question asked | Will they pay? | How do we maximise recovery? |
| Key parameter | PD | LGD and recovery timing |
| Relationship stance | Partner and adviser | Creditor with rights to protect |
| Documents read for | Commercial terms | Remedies, security, events of default, voting thresholds |
| Time horizon | Life of the loan | Months to years of resolution |
| Measured on | Revenue, RAROC | Recovery versus provision, speed |
Multi-creditor dynamics
Most large restructurings involve many creditors: several banks, bondholders, suppliers, credit insurers, tax authorities, employees. They share an interest in preserving value but compete for their share of it. A bank's position depends on its ranking, its security, the voting thresholds in the documentation — majority lenders, usually two-thirds, for most decisions, and all lenders for changes to amounts, dates and pricing — and the procedure that applies if consensus fails. A single holdout lender can block an all-lender decision, which is why modern documentation includes mechanisms to remove or override holdouts in some cases.
Outside the bank, a wider cast appears
The client side
On the client side, the CFO and treasurer negotiate the facility and sign the compliance certificates; the auditors sign off the accounts on which covenants are tested; the company's lawyers negotiate the agreement. In distress, the company appoints financial advisers to design a restructuring and, in France, may seek a court-appointed mandataire ad hoc or conciliateur.
Third parties across the lifecycle
| Third party | Role | Stage |
|---|---|---|
| Rating agencies (S&P Global, Moody's, Fitch) | External ratings used as inputs and early-warning signals | Analysis, monitoring |
| Banque de France | Company cotation, FIBEN database | Analysis, monitoring |
| Valuers | Independent collateral valuation | Structuring, monitoring, workout |
| Credit insurers and guarantors | Share the risk, including export credit agencies | Structuring, workout |
| Data vendors | Financial, market, news and ESG data | Analysis, monitoring |
| Insolvency practitioners (in France, administrateurs and mandataires judiciaires) | Run formal procedures | Workout |
| Distressed-debt investors and credit servicers | Buy or manage non-performing loans | Exit |
Models need owners, validators and supervisors
A model is trusted because it is independently proven
The rating, PD, LGD, EAD and ECL models that run through the lifecycle are themselves governed objects. A model development team builds them from historical data. An independent model validation team tests whether they work: whether they rank borrowers correctly, whether their probabilities match observed default rates, whether they are stable. A model risk committee approves them for use. For IRB models, the supervisor must approve them too, and the ECB has spent years reviewing banks' internal models for consistency. Every override, every recalibration and every change is logged.
Work these through before reading on. Full solutions are in the Answer Key.
3.1 Who owns it?
For each task, name the owner and the independent check: setting the borrower's rating; releasing the first drawdown; deciding that a loan moves to Stage 2; agreeing a covenant reset for a watchlist borrower.
The rating is proposed by the credit analyst and approved or reviewed independently by credit risk, with model validation in the background. The first drawdown is released by the back office after middle office confirms the CPs, under dual authorisation. The move to Stage 2 is determined by the ECL engine's criteria and owned by finance and credit risk, with the provisioning committee as the check on overrides. The covenant reset for a watchlist borrower is negotiated by the RM or the workout unit and approved by the appropriate credit authority, at a higher level because the borrower is on the watchlist.
3.2 Separation
A small bank proposes to save costs by having the same operations team capture new facilities and release payments, with a weekly supervisor review. Identify the risk and propose the minimum control that would make the change acceptable.
One team capturing facilities and releasing payments could create a fictitious or altered facility and pay it out, and a weekly review would find the problem only after the money had left. The minimum acceptable control is that every payment release requires a second authoriser who did not capture the facility, enforced by system entitlements rather than by procedure.
3.3 Holdouts
A syndicated loan's documentation requires two-thirds of lenders for covenant waivers and all lenders for maturity extensions. Lenders holding 80 per cent agree to an amend-and-extend; one lender with 20 per cent refuses. What can and cannot be done, and what options do the majority have?
The majority can waive covenants, because 80 per cent exceeds the two-thirds threshold, but cannot extend maturity, which requires all lenders. Options include buying the holdout's position, offering the extension only to consenting lenders so that the holdout's share matures on the original date, using any replacement or "yank-the-bank" clause in the documentation, or, in a formal procedure, relying on court-supported mechanisms that can bind dissenting creditors.
3.4 Transfer to workout
List three criteria a bank might use to decide when a name moves from coverage and the watchlist to the workout unit, and explain why the transfer itself has value.
Possible criteria: rating at or below a set grade such as grade 8; a payment default or an expected covenant breach without a credible plan; the need for a restructuring or forbearance measure; a defined time on the watchlist without improvement. The transfer has value because it brings specialist skills, a recovery mindset and independence from the original relationship, and because value erodes the longer action is delayed.
Explain to a friend who runs a small business why a bank deliberately makes the person who sells a loan different from the person who approves it, and different again from the person who chases it if it goes wrong. Would they organise their own business the same way?
Why does the bank separate the people who want an outcome from those who control it? Give two examples from the lifecycle.
To prevent conflicts of interest from turning into losses; for example, the RM does not approve the loan, and the person who captures a facility does not release its payment.
What is the difference between the middle office and the back office in loan operations?
The middle office captures and sets up deals, checks CPs and processes amendments; the back office runs fixings, accruals, payments, notices and reconciliations.
What does a provisioning committee do, and why is it needed?
It reviews and approves staging overrides, individual provisions and management overlays, because ECL combines models with judgement on which finance, risk and the business may disagree.
Name three ways in which a workout specialist differs from an originator.
The workout specialist focuses on recovery and LGD rather than PD, treats the client as a counterparty with rights to protect, and reads documents for remedies and voting thresholds.
Who validates a rating model, and why must it be independent of the team that built it?
An independent model validation team, so that the people who built the model are not the ones certifying that it works.
3
Why separate the people who want an outcome from those who control it?
Right. The RM does not approve the loan; the person who captures a facility does not release its payment.
No.
It may slow it, deliberately.
No.
What does the loan back office do?
Credit function.
Coverage.
Right. The middle office captures and sets up deals, checks CPs and processes amendments.
Middle office.
Why is a provisioning committee needed?
No.
Right. It approves staging overrides, individual provisions and overlays.
No.
No.
4Systems, Data, and IT
A loan is a legal promise, but for most of its life the bank knows it only as data. This part follows the facility record through the systems that analyse, approve, book, service, monitor, provision and recover it, and shows where the record is copied, re-keyed and reconciled along the way.
The lifecycle runs on a chain of specialised systems
Ask the clean question first: what does each stage need?
Each stage of the lifecycle needs its own capability, and in most banks each capability lives in a separate system, often from a different vendor and a different decade.
| Stage | System | What it holds | Typical platforms |
|---|---|---|---|
| Origination | CRM | Client, opportunity, deal team | Salesforce, Microsoft Dynamics, in-house |
| Analysis | Spreading tool and rating engine | Spread financials, model scores, overrides, ratings | Moody's CreditLens, S&P tools, in-house |
| Structuring and approval | Loan origination system (credit workflow) | Proposal, routing, decisions, sanction record | nCino, CreditLens, Pega-based, in-house |
| Approval and booking | Limit and exposure system | Approved limits, utilisation, group exposure | In-house, risk platforms |
| Execution | Document management | Facility agreements, CP evidence, legal opinions | Document repositories, LMA-based drafting tools |
| Booking and servicing | Loan servicing system | Facilities, tranches, drawings, schedules, accruals | Finastra Loan IQ, FIS ACBS, core banking |
| Servicing | Payments and general ledger | Cash movements, accounting entries | Payment hubs, SAP, Oracle |
| Monitoring | Covenant tracker and early-warning system | Covenants, tests, headroom; signals and alerts | In-house, vendor modules |
| Monitoring | ECL engine | Staging, scenario-weighted provisions | SAS, Moody's, in-house |
| Workout | Case management and collateral | Watchlist and workout cases, recoveries, valuations | In-house, specialist vendors |
| All | Risk data platform and reporting | Aggregated exposures, regulatory returns | Data warehouse, reporting tools |
The facility record is the thread through all of them
The coverage module showed that the client is the hardest object to keep consistent across systems. In the lifecycle, a second object joins it: the facility. The same facility appears in the credit workflow as a sanction, in the limit system as a limit, in the loan system as a booked commitment, in the covenant tracker as a set of tests, in the ECL engine as an exposure, and in AnaCredit as an instrument. If each system holds its own version, they drift apart, and every report that combines them must reconcile them first.
Spreading tools and rating engines turn documents into grades
Spreading is the most automatable manual step
Spreading — re-casting financial statements into the bank's template — has historically been done by hand: analysts or offshore teams reading PDFs and typing numbers. Modern spreading tools combine optical character recognition with machine-learning extraction that recognises statements, maps line items to the bank's chart of accounts and flags anomalies for a human to check. Banks report that well-implemented automated spreading cuts the time per spread substantially and improves consistency; the condition is that a human still reviews the mapping, because a misclassified line can move a ratio across a covenant or rating threshold.
The rating engine holds the model, the judgement and the trail
The rating engine applies the scorecard to the spread and the qualitative inputs, records the analyst's overrides and their justification, routes the rating for approval and stores the result with a time-stamp. It must be able to reproduce any historical rating exactly: which model version, which inputs, which overrides, who approved. That reproducibility is what supervisors and model validators test.
The loan origination system turns approval into a governed workflow
From an email chain to a digital credit file
A loan origination system, or LOS, holds the credit proposal and its supporting documents, routes it through the delegation engine to the right authority, records each recommendation and decision, and produces the formal sanction record with its conditions. The delegation engine is the delegation matrix turned into executable rules: it reads the deal's exposure, rating, tenor, product and exceptions and routes the proposal accordingly. Because it encodes who may approve what, changes to it are themselves controlled.
The sanction record in the LOS should be the single source for the terms that downstream systems need: the approved limit, the covenants and their thresholds, the conditions precedent and subsequent, the pricing range and the expiry. In practice, many banks still transmit the sanction as a PDF memo, which operations then reads and re-keys. Every re-keyed term is a chance for the leakage described in Part 2, and the reconciliation needed to catch it is a cost that structured data would remove.
Limits and exposure aggregation make authority enforceable
An approval authorises a limit; the limit system enforces it
An approval is only a control if something stops the bank lending more than was approved. That is the job of the limit system. Each approved facility creates a limit, organised in a hierarchy: limits for each facility roll up into a limit for the borrower, which rolls up into a limit for the risk group, which sits within sector, country and portfolio limits derived from the risk appetite. Every drawing consumes the limit at the right levels, and a drawing that would breach a limit is blocked or escalated.
Exposure aggregation is the hardest data problem in the stack
Measuring exposure correctly means adding up, for one risk group, every type of exposure across every product and every legal entity of the bank: drawn loans, undrawn commitments weighted by their conversion factor, guarantees and letters of credit, derivative exposures from hedging, settlement exposures. Each of those lives in a different system with a different client identifier. The large-exposure rules and the coverage module's risk-group hierarchy meet here.
The loan servicing system is the operational heart
What it holds
The loan servicing system holds the facility hierarchy and everything needed to run it: the schedules, the rates, the calendars, the conventions and the history of every event. Its static data — the reference data a loan needs to exist — must come from authoritative sources.
| Static data | Example | Golden source |
|---|---|---|
| Counterparty | Helion Industries SA, LEI, SIREN, role (borrower, guarantor) | Client master data |
| Product | Term loan, revolving facility, guarantee line | Product catalogue |
| Rate index | 3-month EURIBOR, €STR compounded with 5-day lookback | Market data |
| Day-count and business-day convention | Actual/360; modified following | Product and documentation |
| Calendars | TARGET2 calendar for euro; New York for dollar | Market data |
| Settlement instructions | Borrower account, bank's nostro | Standing settlement instructions repository |
| Fee definitions | Commitment fee 35% of margin, quarterly in arrears | Documentation |
The calculation engine is deterministic, and that is its value
At its core, the servicing system is a calculation engine. Given the loan's state, the rate, the day count and the calendar, it computes accruals, interest due, fees and schedules exactly the same way every time. Determinism is what lets operations work by exception: if the inputs are right, the outputs are right, and the team need only look at the breaks.
Rate feeds and fixings
Rates arrive from market-data vendors and are applied to each drawing on its fixing date. For term rates such as EURIBOR, a single fixing per period is needed. For compounded risk-free rates, the system needs every daily rate in the observation period, the lookback convention and the compounding formula, which is why the replacement of LIBOR required most banks to upgrade or replace their loan systems between 2020 and 2024.
Accounting: amortised cost and the effective interest rate
Loans held to collect their contractual cash flows are accounted for at amortised cost under IFRS 9. Fees that are an integral part of the loan's yield — notably arrangement fees — are not recognised as income on the day they are received; they are spread over the life of the loan through the effective interest rate, a single rate that discounts all the loan's expected cash flows, fees included, to its initial carrying amount.
Covenants must become data to be monitored
The free-text problem
Covenants are written in prose in a long agreement, with definitions that run to pages. To be monitored, each must become a structured record: what is tested, how each term is defined, the threshold and how it steps over time, the test dates, the source of the figures, the remedy period, who must be notified. The hardest IT challenge in covenant monitoring is therefore at the very start: capture.
| Field | Helion leverage covenant |
|---|---|
| Covenant type | Financial, maintenance |
| Ratio | Net debt / EBITDA |
| Definitions | As defined in clause 22.1; leases included in debt; pro-forma EBITDA for acquisitions |
| Threshold schedule | 3.25x for first two test dates after completion, 3.50x thereafter |
| Test frequency and basis | Semi-annual, rolling twelve months |
| Information source | Compliance certificate within 120 days of year end and 90 days of half year |
| Cure rights | None |
| Remedy on breach | Event of default after notice |
| Linked sanction condition | Credit approval ref. 2026-GC-0413, condition 2 |
Machine reading of agreements
Natural-language processing and large language models are now used to extract covenants, definitions and key terms from facility agreements into structured form. The value is large — consistent capture across thousands of agreements, and lineage from each data field back to the clause it came from — and so is the need for human review, because a misread definition is a mis-monitored covenant. Leading implementations present the extracted value side by side with the source clause for a reviewer to confirm.
Testing engines and continuous headroom
Once covenants are data, a testing engine can recompute them from the borrower's spread financials rather than relying on the compliance certificate, compare the bank's figure with the borrower's, track headroom over time and raise alerts when headroom falls through thresholds. The frontier is to compute headroom continuously from the latest available data — management accounts, payment flows — rather than twice a year.
Early-warning systems and ECL engines close the monitoring loop
Early-warning systems: signals in, scores out, alerts actioned
An early-warning system ingests behavioural, financial, market and qualitative signals for every borrower, combines them into a score or a set of triggers, and routes alerts to the RM and the credit monitoring team with a deadline to respond. Its value depends less on the sophistication of the model than on the discipline of the response: an alert that is acknowledged but not acted on is worse than no alert, because it creates a record that the bank knew.
The ECL engine stages and provisions every exposure
The ECL engine is the industrial calculator of the monitoring stage. Every reporting date, for every exposure, it takes the booked balance and schedule from the loan system, the rating and PD term structure from the rating engine, the LGD and EAD parameters, the forward-looking macroeconomic scenarios and their weights, and the staging criteria. It assigns each exposure to Stage 1, 2 or 3 and computes the discounted, probability-weighted expected credit loss. Its output flows to the general ledger as the provision, to financial reporting, and to regulatory reporting.
Management overlays and the audit trail
Models never capture everything. When the bank believes the model under- or overstates risk — for a sector shock the historical data has never seen, for instance — it applies a management overlay, an adjustment approved by the provisioning committee. Overlays became widespread during the pandemic and the 2022 energy shock, and supervisors scrutinise them closely: each must be justified, quantified, approved and tracked until it is released.
Workout systems manage cases, collateral and recoveries
From spreadsheets to governed cases
Workout has long been the least automated part of the lifecycle, run on emails and spreadsheets by specialists. Modern case-management systems give each distressed name a case with a strategy, tasks, deadlines, documents, committee decisions and an audit trail. Collateral systems hold what the bank can seize — assets, their legal status, their valuations and the date of the last valuation — and recovery tracking records what the bank expects to recover and what it actually collects, which feeds both the Stage 3 provision and the bank's LGD models.
Regulatory reporting draws on every stage
Loan-level and aggregate returns
| Return | What it requires | Data sources |
|---|---|---|
| AnaCredit | Loan-by-loan data on credit to legal entities in the euro area above 25,000 euros, monthly to the national central bank and the ECB | Loan system, client master data, collateral, risk parameters |
| COREP | Capital requirements, including credit risk by exposure class and approach | Exposure, ratings, risk parameters |
| FINREP | Financial reporting, including performing and non-performing exposures, forbearance, provisions | Ledger, ECL engine, forbearance flags |
| Large exposures | Exposures to clients and connected groups above thresholds | Exposure aggregation, risk groups |
| NPL and forbearance templates | Detailed data on non-performing and forborne exposures | Workout, collateral, ECL |
AnaCredit illustrates why the facility record matters. It requires dozens of attributes per instrument and per counterparty — identifiers, amounts, dates, interest-rate type, collateral, probability of default, accounting stage — drawn from the CRM, the loan system, the collateral system and the risk engines, and cross-checked by the central bank. Inconsistencies between those sources come back as data-quality queries.
Integration: where the lifecycle leaks
The re-keying problem
Follow one term — Helion's leverage covenant threshold of 3.25x — through a typical bank. It is typed into the credit proposal, discussed in committee, written in the sanction memo, negotiated into the facility agreement, keyed into the loan system's facility notes, keyed again into the covenant tracker and keyed a third time into the monitoring team's spreadsheet. Seven appearances, four of them manual, and nothing ensures they agree.
What a better architecture looks like
| Pattern | Today | Target |
|---|---|---|
| Sanction to booking | PDF memo, re-keyed | Structured sanction record passed by API to booking |
| Agreement to systems | Read and keyed by operations | Extracted as structured data with clause-level lineage, reviewed once |
| Client and facility identity | Different identifiers per system | Shared client and facility identifiers from master data |
| Exposure | Nightly batch aggregation | Event-driven updates on every drawing and repayment |
| Covenants | Spreadsheet calendar | Tracker fed by the agreement data and the spread financials |
| Reconciliation | Front-to-back-to-ledger breaks investigated manually | Continuous automated reconciliation; humans handle exceptions |
Work these through before reading on. Full solutions are in the Answer Key.
4.1 Follow the data
List every system that must change, and in what order, when Helion repays 15 million euros of Facility A on its first anniversary.
The loan servicing system records the repayment against the schedule and reduces the outstanding; payments collect the 15 million and reconciliation matches it; the general ledger receives the entries; the limit system reduces the term-loan limit, since a term loan cannot be redrawn; exposure aggregation and the risk data platform update the group exposure; the covenant calculation reflects lower net debt at the next test; the ECL engine uses the lower EAD at the next reporting date; and AnaCredit reports the new outstanding at the next monthly submission.
4.2 Static data
A revolving facility in dollars is booked with the TARGET2 calendar instead of the New York calendar. Describe two concrete errors this will cause.
Payment and rate-fixing dates will be set on days that are business days in the euro area but holidays in New York, so payments may fail or be late. Interest periods adjusted to the wrong business days will contain the wrong number of days, so interest will be miscalculated.
4.3 Fee recognition
A 300 million euro, three-year bullet term loan carries a 0.60 per cent arrangement fee. How much cash does the bank receive at signing, and approximately how much fee income does it recognise in the first year under the effective-interest method?
The bank receives 0.60% × 300 million = 1.8 million euros in cash at signing. On a three-year bullet loan, the effective-interest method spreads the fee roughly evenly, so about 0.6 million is recognised in the first year.
4.4 Exposure
A borrower has a 50 million euro term loan fully drawn, an 80 million revolving facility 30 million drawn, 20 million of issued guarantees and 10 million of derivative exposure. What is the total exposure for limit purposes if all commitments count in full, and what is the EAD for capital purposes if undrawn commitments carry a 40 per cent conversion factor and guarantees a 50 per cent factor?
For limits: 50 + 80 + 20 + 10 = 160 million euros. For capital: 50 (term loan) + 30 drawn + 40% × 50 undrawn = 20 + 50% × 20 guarantees = 10 + 10 derivatives = 120 million euros.
4.5 Design a covenant record
Write the structured record for an interest-cover covenant of at least 4.0x, tested quarterly on a rolling twelve-month basis, with a 30-day equity-cure right for the sponsor.
Type: financial, maintenance. Ratio: EBITDA / net finance charges, as defined in the agreement. Threshold: not below 4.0x. Frequency: quarterly, rolling twelve months. Source: quarterly compliance certificate and management accounts within an agreed number of days of quarter end. Cure: equity cure by the sponsor within 30 days of the certificate date, with the cure amount added to EBITDA or deducted from debt as the agreement defines and limits on the number of cures. Remedy: event of default if uncured. Links: sanction reference and clause reference.
Explain to someone who has changed banks why a corporate loan agreement of two hundred pages still has to be read and typed into systems by people, and why that matters to whether the bank notices when the borrower gets into trouble.
What does the delegation engine in a loan origination system do?
It applies the delegation matrix to each proposal's attributes to route it to the right authority, and records who decided what, when.
Why is exposure aggregation described as the hardest data problem in the stack?
Because it requires every type of exposure, across every product and legal entity, to be mapped to the right risk group despite different client identifiers in each system.
Name four items of static data a loan servicing system needs and their golden sources.
For example: counterparty from client master data; rate index and calendars from market data; day-count convention and fee definitions from the documentation and product catalogue; settlement instructions from the standing-instructions repository.
Why is an arrangement fee not recognised as income on the day it is received?
Because under IFRS 9 fees integral to the loan's yield are part of the effective interest rate and recognised over the loan's life.
Why is covenant capture the hardest step in covenant monitoring?
Because covenants and their definitions are written in prose across long documents and must be turned into precise structured data before anything can be tested.
What does the ECL engine need from other systems at each reporting date?
Balances and schedules from the loan system, ratings and PD term structures, LGD and EAD parameters, staging inputs such as watchlist and forbearance flags, and macroeconomic scenarios and weights.
4
What does a delegation engine do?
Right.
Payments.
Pricing.
Servicing.
Why is exposure aggregation the hardest data problem?
No.
Right.
No.
It is required.
Where should a loan's rate index and calendars come from?
No.
That is the error the golden source avoids.
Right. Counterparty data comes from client master data; conventions from documentation.
Not a golden source.
5Risks, Pain Points, and Transformation
The lifecycle is long, handled by many people and carried by many systems. This part maps where it fails, looks at three real corporate collapses through the eyes of their lenders, and sets out how banks are rebuilding the lifecycle to be faster without becoming riskier.
A MECE map of where the lifecycle fails
One failure mode per stage
Every stage has a characteristic way of going wrong. Taken together, the failure modes below cover the field without overlapping.
| Stage | Characteristic failure | Typical consequence |
|---|---|---|
| Origination | Deal pre-sold to the client; credit brought in late | Compressed analysis, pressure to approve |
| Credit analysis | Wrong data, optimistic projections, negotiated ratings | Mispriced risk, wrong capital and provisions |
| Structuring | Covenants too loose or definitions too generous | No seat at the table until too late |
| Approval | Weak challenge; decisions outside delegated authority | Concentration, policy breaches, supervisory findings |
| Execution and booking | Leakage between sanction, agreement and system; CPs missed | Unprotected exposure, wrong servicing |
| Servicing | Calculation and static-data errors; unreconciled breaks | Client overcharging, financial loss, conduct remediation |
| Monitoring | Late detection; alerts not acted on; untested covenants | Higher losses, late provisions, procyclical shocks |
| Workout and exit | Late transfer; stale collateral data; poor recovery estimates | Lower recovery, misstated provisions, phantom exposures |
Two roots again
As in coverage, most of these failures share two roots. The first is fragmented data: the same client and facility held in different forms in different systems, with manual copying between them. The second is delayed judgement: decisions taken too late because information arrived too late, or because the incentive to act was weaker than the incentive to wait.
Procyclicality: when risk systems deepen the downturn
The self-reinforcing loop
In a downturn, borrowers are downgraded, loans migrate to Stage 2, provisions jump, capital requirements rise with the higher PDs, profits fall and capital ratios weaken. Banks respond by lending less and pricing higher, which hurts borrowers further and deepens the downturn. IFRS 9 made this loop faster, because the move to lifetime provisions happens as soon as risk increases significantly rather than when losses occur.
The dampeners
Several mechanisms limit the loop. Through-the-cycle rating philosophies for capital, the countercyclical capital buffer that supervisors can release in a crisis, the weighting of several macroeconomic scenarios in ECL rather than a single pessimistic one, and supervisory guidance — as during the 2020 pandemic, when supervisors encouraged banks not to treat public support measures such as payment moratoria as automatic triggers for Stage 2. The design question for every bank is how to make provisions responsive to real deterioration without making them amplify every economic wobble.
Three real failures, seen from the lenders' side
Each case illustrates a different stage
The cases below are well documented in public sources. Each is told from the perspective of the lenders, and each illustrates a different weakness in the lifecycle.
Carillion, 2018 — when the definitions hid the debt
What happened
Carillion, one of the United Kingdom's largest construction and public-services contractors, issued a profit warning in July 2017 with a provision of 845 million pounds against its contracts. Further write-downs followed, its banks declined to provide more funding, and in January 2018 it went into compulsory liquidation, with tens of thousands of employees and a long chain of unpaid subcontractors affected.
Root cause
Carillion's reported earnings relied on aggressive contract accounting, and its cash position was supported by an early-payment programme for suppliers — a form of supply-chain finance in which banks paid suppliers early and Carillion repaid the banks later. The amounts owed under this programme, reported to have reached several hundred million pounds, were presented within payables rather than as borrowings. Measured on reported figures, leverage looked manageable; measured with the supplier-finance liability treated as debt, it did not.
The lesson
Covenant definitions decide the answer. A leverage covenant that does not capture supplier finance, factoring or other debt-like obligations can be met by a company that is in fact highly indebted. Credit analysis must look through the presentation to the economics, and structuring must write definitions that do the same. Supervisory and accounting attention to supplier-finance disclosure has since increased, with new IFRS disclosure requirements applying from 2024.
Wirecard, 2020 — when the data itself was false
What happened
Wirecard, a German payments company and member of the DAX index, announced in June 2020 that its auditor could not confirm the existence of about 1.9 billion euros of cash supposedly held in trust accounts in Asia. Days later the company admitted that the cash probably did not exist, and it filed for insolvency. A syndicate of banks had provided Wirecard with a revolving credit facility of about 1.75 billion euros.
Root cause
The lenders, like the market, relied on audited financial statements that turned out to be fraudulent. Warning signs had been public for some time — investigative journalism questioning the Asian business, short-seller reports, delays in the audit — but they were contested by the company and, for a time, by its supervisor.
The lesson
Monitoring cannot stop at the compliance certificate and the audited accounts. Early-warning systems that include adverse media and market signals, and credit analysts empowered to treat persistent, credible allegations as a reason to dig deeper rather than wait for proof, are the bank's defence against the rare but devastating case in which the data is false. Liquidity and cash verification deserve particular scepticism when cash sits with third parties in opaque structures.
Steinhoff, 2017 onwards — when many lenders must act together
What happened
In December 2017, Steinhoff International, a global retail group listed in Frankfurt and Johannesburg, announced accounting irregularities. Its chief executive resigned, its share price collapsed and its access to funding disappeared. The group had raised billions of euros from banks and bond investors through several entities.
Root cause
The immediate cause was fraud. The lending lesson lies in what followed: a complex, multi-jurisdictional group with debt at several levels, owed to many creditors with different rankings and rights, had to be stabilised before any enforcement could destroy the value left. Lenders agreed to standstills and a lock-up, and a restructuring of the group's European debt was implemented over the following years through a series of court-supported procedures.
The lesson
In a large default, the bank's recovery depends less on its own documentation than on the collective behaviour of creditors and the procedures available to bind them. Knowing exactly which entity owes what to whom — the legal hierarchy and the risk group of the coverage module — is the starting point, and the standstill, the most basic tool in Part 2, is often what preserves the going-concern value long enough for a solution.
Six transformation themes are rebuilding the lifecycle
Speed and control together
| Theme | What it means | Stage |
|---|---|---|
| The tiered credit factory | Route standard, low-risk credits through automated, fast lanes and reserve human judgement for complex deals | Analysis, approval |
| Automated spreading and AI credit memos | Machine extraction of financials; generative AI drafts of the credit proposal from structured data | Analysis |
| Structured sanctions and digital documentation | Sanction and agreement terms captured once as data, with clause-level lineage, feeding booking and monitoring | Approval, execution |
| Exception-based servicing | Straight-through processing for standard events; people handle only the breaks | Servicing |
| Continuous monitoring | Real-time early warning, covenant headroom computed from live data, event-driven re-rating | Monitoring |
| Digitised workout and NPL markets | Case management, collateral data, analytics on recoveries, deeper secondary markets | Workout |
The tiered credit factory
The credit factory squares speed with control by matching the intensity of governance to the risk of the decision. A small, well-rated facility within policy can be assessed by a scorecard, approved by a single credit officer with system checks, and documented on a standard template in days. A large, complex or policy-exception deal receives full analysis, committee debate and bespoke documentation. The gain comes from not treating every credit as if it were the hardest one, while proving through data and controls that the fast lane is not a lower standard.
Generative AI in credit, with a named human on every decision
Generative AI is being used to draft credit proposals from spread data and the client file, summarise annual reports and news, compare a signed agreement against the sanction, extract covenants from documents, and prepare workout case summaries. The pattern that works keeps the model as a drafter and the analyst or officer as the author and decision-maker. Outputs are grounded in the bank's own data, reviewed and logged, and the decision remains attributable to a named person.
Under the EU AI Act, AI systems used to evaluate the creditworthiness of natural persons are classified as high-risk, with obligations on risk management, data governance, human oversight and transparency. Corporate credit assessment is not in that category as such, but credit to small businesses often involves assessing the owner personally, and banks' own model-risk frameworks apply in any case. Supervisors also expect machine-learning rating models used for capital purposes to meet the same standards of explainability and validation as traditional scorecards, which has slowed their adoption in IRB models even where their predictive power is higher.
ESG and climate risk enter every stage
The EBA guidelines on loan origination and monitoring expect banks to consider ESG factors when assessing borrowers, and the ECB expects banks to integrate climate-related and environmental risks into credit risk management. In practice this means an ESG assessment in the credit proposal, transition and physical climate risk in sector policies and ratings, ESG terms such as sustainability-linked margins in structuring and servicing, and ESG data in the monitoring and reporting chain. The coverage module described the product side; here the point is that ESG adds a new data domain to every stage of the lifecycle.
The target operating model is one governed data spine
Three phases, as in coverage
The target operating model for the lifecycle mirrors the coverage roadmap. The first phase builds the data spine: shared client and facility identifiers, a structured sanction record, clause-level extraction of agreements, and automated reconciliation between the loan system, the limit system and the ledger. The second phase redesigns the processes on that spine: the tiered credit factory, exception-based servicing, covenant monitoring fed by the agreement data. The third phase adds intelligence: AI-assisted analysis and drafting, continuous early warning, predictive recovery analytics.
Measuring whether it worked
| Metric | What it measures | Direction |
|---|---|---|
| Time to decision | Days from complete proposal to sanction, by tier | Down, especially in the fast lane |
| Time to cash | Days from signing to first drawdown | Down |
| Re-keying count | Manual entries of the same term across systems | Down, towards one |
| Servicing breaks | Unreconciled items per thousand events | Down |
| Covenant capture accuracy | Share of covenants captured correctly first time | Up |
| Early detection | Share of defaults that were on the watchlist six months before | Up |
| Recovery rate | Recoveries as a share of exposure at default | Up |
Work these through before reading on. Full solutions are in the Answer Key.
5.1 Place the failure
For each incident, name the stage and the root cause: a covenant approved at 3.0x is set up in the tracker at 3.5x; an RM ignores three early-warning alerts on a client because a refinancing is being negotiated; interest on a dollar loan is calculated with the wrong calendar for two years; a workout team discovers that a key guarantee was never signed.
The covenant set up at 3.5x is an execution and booking failure — leakage between sanction and system — rooted in re-keyed data. The ignored alerts are a monitoring failure rooted in delayed judgement and conflicting incentives. The wrong calendar is a servicing failure rooted in static data. The unsigned guarantee is an execution failure — a CP not checked — discovered in workout, rooted in weak CP control.
5.2 Definitions
A company has EBITDA of 300 million euros, financial debt of 900 million, cash of 100 million and 250 million owed under a supplier-finance programme reported in trade payables. Compute leverage with and without the supplier-finance liability, and say which a lender should use and why.
Without supplier finance: (900 − 100) / 300 = 2.7x. With it: (900 + 250 − 100) / 300 = 3.5x. A lender should use the figure that includes supplier finance, because the amount is owed to banks and behaves like debt: if the programme is withdrawn, the company must find the cash.
5.3 Design the fast lane
Propose four eligibility criteria for a credit-factory fast lane and two controls that prove it is not a lower standard.
Possible criteria: total group exposure below a set amount; rating in the investment-grade band; no policy exceptions; standard documentation and product; no watchlist or adverse KYC flags. Controls: automated eligibility checks that cannot be overridden by the front office, and periodic back-testing of fast-lane losses and sample reviews by credit risk and internal audit.
5.4 Sequence the programme
A bank wants to deploy a generative-AI tool that drafts credit proposals. Its spread financials are held in spreadsheets, and its sanctions are PDF memos. Advise the sponsor in three sentences.
The tool can draft only from the data it can read, and today the inputs are spreadsheets and PDFs, so it would draft confidently from inconsistent sources. Start by moving spreading into a structured tool and sanctions into structured records for a pilot segment. Launch the drafting tool on that segment, measure analyst time saved and error rates, and extend it as the data spine extends.
5.5 Carry it forward
Name one idea from each of Parts 1, 2 and 4 that would have helped a lender in one of the three case studies.
From Part 1: separating the first and second ways out — Wirecard's lenders relied on a cash position that did not exist. From Part 2: covenant definitions and early-warning signals — Carillion's supplier finance and Wirecard's adverse media. From Part 4: exposure aggregation and the legal hierarchy — essential to organise creditors in Steinhoff.
Explain to a family member why a bank could lend to a famous, listed company whose accounts were signed by a major auditor and still lose most of its money. Use Wirecard, and avoid the words "covenant", "syndicate" and "audit opinion".
What are the two roots of most lifecycle failures?
Fragmented data and delayed judgement.
What is procyclicality, and name two mechanisms that dampen it.
The tendency of risk systems to amplify downturns by raising provisions and capital requirements as conditions worsen; dampeners include through-the-cycle capital models, the countercyclical buffer, scenario-weighted ECL and supervisory flexibility.
Which stage does each of the three case studies illustrate, and what is its lesson?
Carillion: structuring and analysis, definitions must capture all debt-like obligations. Wirecard: monitoring, signals beyond audited data matter. Steinhoff: workout, collective creditor action preserves value.
What is a tiered credit factory, and why does it not lower standards?
A model that routes credits to different levels of process according to risk and complexity; it does not lower standards because eligibility is controlled by system rules and outcomes are back-tested and audited.
Why must the data spine come before AI in a lending transformation?
Because AI and new workflows draw on the facility record; if that record is fragmented and re-keyed, they make the errors faster rather than removing them.
5
What are the two roots of most lifecycle failures?
No.
No.
No.
Right.
Which case illustrates the monitoring stage?
Right. Carillion is structuring and analysis; Steinhoff is workout.
Wirecard is.
Workout and collective creditor action.
Structuring and analysis.
What is procyclicality?
No.
Right. Through-the-cycle models and buffers dampen it.
Too narrow.
No.
Module Summary — The Six Ideas That Matter Most
If you retain only a handful of things, retain these. Each compresses a part of the module into a single, load-bearing idea.
A loan is a promise with two ways out, measured by three numbers
The first way out is the borrower's cash flow, measured by the probability of default; the second is what the bank can recover, measured by the loss given default. Exposure at default completes the set, and for a revolving facility it exceeds the drawn balance because troubled borrowers draw their lines. Expected loss, PD × LGD × EAD, is a cost covered by pricing and provisions; unexpected loss is covered by capital. The same parameters therefore drive the price, the capital and the provisions, and an error in one is an error in all three.
Deciding takes weeks; living with the loan takes years
The lifecycle runs through eight stages in three phases. Origination, analysis, structuring and approval decide whether and how to lend. Execution and booking put the loan in place. Servicing, monitoring and workout or exit fill the rest of its life. Most losses come from loans that were reasonable decisions and deteriorated later, so the long middle of the lifecycle decides what the decision was worth.
Structure and governance turn a risky borrower into an acceptable loan
Amount, tenor, amortisation, pricing, security and covenants are the levers that fit a loan to the borrower's need and the bank's appetite. Covenants are not traps but early seats at the table, and their definitions decide what they actually measure. Approval is a governed, delegated, four-eyes decision, usually conditional, and the sanction is the source of truth that the documentation and the systems must match.
Booking and servicing are where the contract becomes data
A signed agreement becomes a facility–tranche–drawing record that feeds servicing, payments, risk, finance, reporting and monitoring. Interest is rate times balance times time under a day-count convention, fees run on their own calendars, and every movement must reconcile. Errors here are errors everywhere downstream, and the cost of re-keying the same terms into several systems is paid in leakage and reconciliation.
Monitoring decides the losses, and IFRS 9 makes deterioration visible fast
Covenant headroom, early-warning signals and event-driven re-rating detect trouble; the significant-increase-in-credit-risk test moves a loan from twelve-month to lifetime provisions, which can multiply the provision without a missed payment. Default, forbearance and non-performing status follow harmonised European definitions. In workout, the question changes from "will they pay?" to "how do we maximise recovery?", and early action preserves the most value.
The lifecycle is a data pipeline, and transformation removes its seams
Every stage runs on a different system holding its own copy of the client and the facility. Control leaks wherever data is re-keyed. The credit factory, automated spreading, AI drafting, structured sanctions and documents, exception-based servicing and continuous monitoring all deliver only when built on a single, governed facility record, which is why the data spine comes first.
Process Cheat Sheet — The Lifecycle on One Page
A single reference for the eight stages. Everything below is developed in full in Parts 2 to 4.
| Stage | Core activity | Key output | Lead | Core system |
|---|---|---|---|---|
| 1. Origination | Opportunity, KYC check, early credit indication | Logged opportunity, deal team | RM | CRM |
| 2. Credit analysis and rating | Data, spreading, business and financial risk, rating, proposal | Rating and PD, credit proposal | Credit analyst | Spreading tool, rating engine |
| 3. Structuring | Amount, tenor, amortisation, pricing, security, covenants | Term sheet, RAROC | Structurer | Pricing tools |
| 4. Approval | Delegation routing, independent review, committee | Conditional sanction, approved limit | Credit officer, committee | Loan origination system |
| 5. Execution and booking | Documentation, CPs, capture, set-up, limit activation, first drawdown | Signed agreement, booked facility | Legal, middle office | Document management, loan system, limits |
| 6. Servicing | Rate fixing, accrual, fees, drawdowns, repayments, notices, reconciliation | Correct cash flows and records | Back office, agent | Loan system, payments, ledger |
| 7. Monitoring | Covenants, early warning, re-rating, staging, provisions, portfolio review | Current rating, ECL, watchlist decisions | Credit risk, RM, finance | Covenant tracker, EWS, ECL engine |
| 8. Workout and exit | Repayment or refinancing; or watchlist, restructuring, sale, enforcement, write-off | Resolved exposure | RM; workout unit | Case management, collateral, recovery |
Key numbers to remember
| Metric | Rule of thumb |
|---|---|
| Expected loss | PD × LGD × EAD |
| Senior unsecured corporate LGD | About 40–45 per cent |
| Credit conversion factor, committed undrawn (standardised, from 2025) | Commonly 40 per cent; 10 per cent if unconditionally cancellable |
| Regulatory PD floor for corporates | 0.05 per cent |
| Output floor, fully phased in | 72.5 per cent of standardised RWA |
| Large-exposure limit | 25 per cent of Tier 1 capital per client or connected group |
| Default definition | More than 90 days past due on a material amount, or unlikely to pay |
| Corporate materiality threshold | 500 euros and 1 per cent of exposure |
| IFRS 9 SICR backstop | 30 days past due |
| Forbearance probation, performing | At least 2 years |
| Euro day count | Actual/360 |
| Commitment fee | Often 35 per cent of the margin |
| NPL backstop, unsecured | Full coverage after 3 years |
The one-sentence version
A corporate loan is a promise that the bank analyses, shapes, approves and documents in a few weeks, then has to service, watch and, if needed, rescue for years, through a chain of people and systems that work only as well as the single facility record they all depend on.
Sources and further reading
Regulatory positions and market facts stated in this module reflect sources available in mid-2026. Regulation in this field changes on a scale of months, so treat every date below as a fact with a shelf life and verify it before relying on it in client work.
How to read this section
Where the module states a legal or accounting obligation, it names a specific instrument, given below so it can be checked at source. Figures explicitly flagged as illustrative — the Helion case throughout, the master scale, the delegation matrix and the pricing, provisioning and recovery examples — are invented to teach a mechanic and have no source, by design.
Primary legislation and standards referenced
| Short name used in the module | Instrument | Relevance |
|---|---|---|
| CRR and CRR3 | Regulation (EU) No 575/2013, as amended by Regulation (EU) 2024/1623 | Capital, IRB, credit conversion factors, output floor, large exposures, definition of default |
| NPL backstop | Regulation (EU) 2019/630 amending the CRR | Minimum loss coverage for non-performing exposures |
| IFRS 9 | IFRS 9 Financial Instruments, as adopted in the EU | Classification, effective interest rate, staging and ECL |
| EBA loan origination and monitoring | EBA/GL/2020/06 | Credit granting, assessment, pricing, collateral valuation, monitoring |
| EBA definition of default | EBA/GL/2016/07 and the related RTS on materiality | Days past due, materiality thresholds, unlikeliness to pay |
| AnaCredit | Regulation (EU) 2016/867 of the ECB | Granular credit data reporting |
| Credit servicers directive | Directive (EU) 2021/2167 | Sale and servicing of non-performing loans |
| Restructuring directive | Directive (EU) 2019/1023, transposed in France by Ordinance No 2021-1193 | Preventive restructuring, classes of creditors, cross-class cram-down |
| AI Act | Regulation (EU) 2024/1689 | High-risk classification of creditworthiness assessment of natural persons |
Dated facts and where they come from
IFRS 9 has applied to annual periods beginning on or after 1 January 2018. The EBA guidelines on loan origination and monitoring applied to loans originated from 30 June 2021. The EBA definition-of-default guidelines applied from 1 January 2021. The final Basel III reforms began applying in the EU on 1 January 2025, with the output floor phased in towards 72.5 per cent.
LIBOR: sterling, yen, Swiss franc and euro LIBOR settings ceased at the end of 2021; the US dollar panel settings ceased after 30 June 2023, and the remaining synthetic settings had ended by the end of September 2024. €STR has been published by the ECB since October 2019.
Amendments to IAS 7 and IFRS 7 requiring disclosure of supplier finance arrangements apply to annual periods beginning on or after 1 January 2024.
Carillion entered compulsory liquidation in January 2018 after a July 2017 profit warning that included an 845 million pound contract provision; the UK parliamentary committees' joint report of May 2018 describes the use of its early-payment facility. Wirecard filed for insolvency in June 2020 after announcing that about 1.9 billion euros of cash could not be confirmed; its bank lenders had provided a revolving credit facility of about 1.75 billion euros. Steinhoff announced accounting irregularities in December 2017; its debt restructuring proceeded over the following years through court-supported procedures. These facts are drawn from public reporting and official reports and should be checked against the primary sources before reuse.
Where to go deeper
For capital and provisioning, the EBA's interactive single rulebook and the IASB's IFRS 9 materials are authoritative; the ECB's supervisory guidance on IFRS 9, NPLs and climate risk is the most practical. For documentation, the Loan Market Association's guides to its recommended forms explain the clauses behind every stage of the lifecycle. For servicing conventions after LIBOR, the working groups on euro and sterling risk-free rates published detailed recommendations on compounding, lookbacks and fallbacks. For workout and French procedures, the Commercial Code's Book VI and the commentaries following the 2021 reform set out the procedures; the UK parliamentary report on Carillion and the published investigations into Wirecard are instructive case material.