Invordering
Invordering is kontant wat op wanbetalings- of afgeskryfde lenings ingevorder word. Leer hoe invorderingskoerse bereken word, wat hulle aandryf, en hoe hulle met LGD en ECL verband hou.
Recovery is cash collected on a loan after it has defaulted — and in the narrower, more common usage, cash collected on a loan that has already been written off and removed from the balance sheet.
The term is used in two related senses, and they should be kept distinct:
- Recovery on defaulted exposure still on the books — amounts collected on a non-performing loan before write-off, including collateral realisation and guarantor payments.
- Post-write-off recovery — amounts collected on accounts already derecognised. This is what "recoveries" usually means in a lender's management reporting and in the structure of its collections organisation.
A write-off is an accounting judgement that recovery is no longer reasonably expected. It is not a legal waiver: unless the lender has formally forgiven the debt, the claim survives derecognition and pursuit may continue. Recovery is what happens in the space between those two facts.
Recovery vs Collections
CollectionsRecoveryAccount statusDelinquent, still on balance sheetDefaulted or written offPrimary objectiveCure — return the account to currentCash, at acceptable costRelationshipPreserved where possibleGenerally overSuccess rateHigh in early bucketsLow, and declining with ageTypical modelIn-house, relationship-basedIn-house specialist, external agency, or legalAccounting effectReduces arrearsRecognised in profit or loss
The operating logic differs fundamentally. Collections is an exercise in diagnosing why a borrower missed and removing the obstacle. Recovery is an exercise in economics: what can realistically be obtained, at what cost, over what timeframe.
How Recovery Rate Is Calculated
Period-based recovery rate
Recovery rate = Recoveries received in the period / Amounts written off in the period × 100
This is the simplest form and the most commonly quoted, but it is structurally unreliable: the numerator relates to loans written off in earlier periods while the denominator relates to loans written off in this one. Where write-off volume is lumpy — as it is wherever write-offs are approved periodically rather than processed continuously — the ratio swings wildly with no change in underlying performance.
Use it only for rough period tracking, and never for comparison.
Cohort (vintage) recovery rate
Cohort recovery rate = Cumulative recoveries on a write-off cohort / Original written-off balance of that cohort × 100
Track a defined write-off cohort — all loans written off in a given month or quarter — and measure cumulative recovery against its original balance over time. This is the only method that produces comparable, decision-useful figures.
Recovery rate and loss given default
LGD = 1 − Recovery rate
Recovery rate is the direct complement of loss given default, the parameter used in expected credit loss calculation. This is the link that makes recovery measurement matter beyond the collections department: the LGD assumption feeding the ECL model should be grounded in the institution's own observed cohort recovery experience, not in a benchmark or a guess.
Crucially, the LGD-relevant recovery rate must be economic, not nominal: net of the costs of obtaining recovery, and discounted for the time taken to receive it.
The Recovery Curve
Recovery is heavily front-loaded and then trails off. A typical cohort behaves like this:
Period after write-offRecoveriesCumulative recovery rate0–6 months60,0006.0%7–12 months45,00010.5%13–24 months35,00014.0%25–36 months12,00015.2%37+ months8,00016.0%
Illustrative, on a written-off cohort of 1,000,000.
Two implications follow:
Speed matters more than persistence. Roughly two-thirds of what will ever be recovered arrives in the first year. Delays in placing accounts with a recovery unit or agency permanently reduce the total, they do not merely postpone it.
Nominal recovery overstates economic recovery. Take the 16% cumulative figure above, apply a 25% contingency fee to agency-collected amounts and discount the cash flows at the institution's cost of capital, and the economic recovery rate can easily fall to single digits. The nominal figure is what gets reported; the economic figure is what should feed LGD and any decision about whether to pursue at all.
Sources of Recovery
- Voluntary borrower payment — circumstances improve, or the borrower wants access to credit again. Credit bureau listing is often the practical driver here.
- Negotiated settlement — accepting a discounted lump sum to close the account. Frequently the highest-value option, since a partial payment now usually beats a full claim never collected.
- Guarantor claim — where a personal guarantee exists and the guarantor has capacity.
- Group fund or joint liability — in group lending, amounts contributed by fellow members.
- Compulsory savings offset — applying the borrower's own savings balance against the debt, where the loan agreement and regulation permit.
- Collateral realisation — repossession and sale of pledged assets, net of seizure, storage and disposal costs.
- Insurance — credit life cover on borrower death, crop or livestock insurance, or portfolio guarantee schemes.
- Legal enforcement — judgment, attachment of assets, garnishment of wages or bank accounts.
- Sale of the debt — assignment of a written-off portfolio to a third-party purchaser at a discount to face value.
A measurement caution: recovery from the borrower's own compulsory savings, from a group fund, or from insurance is not evidence of borrower repayment capacity. Aggregating these with voluntary borrower payment inflates the apparent recovery rate and corrupts any LGD estimate built from it. Track the sources separately.
Recovery Routes and Their Economics
In-house recovery unit. Full control over conduct and data, and lower marginal cost where volumes justify dedicated staff. Requires different skills and incentives from origination or early collections.
External agency. Usually contingency-based, at a fee proportional to amounts recovered. Cost is variable and aligned, but the institution retains regulatory and reputational responsibility for agent conduct — outsourcing the activity never outsources the accountability. Contract terms should bind agents to the institution's own conduct standards, with monitoring.
Legal action. Appropriate only where the exposure is large enough to justify it. The relevant test is the expected recovery, discounted and net of legal cost, weighted by the probability of both winning and actually enforcing — enforcement being where most judgments fail. Court timelines and enforcement mechanics vary enormously by jurisdiction, and a judgment against a borrower with no attachable assets has no value.
Portfolio sale. Selling a written-off book converts an uncertain long-tail cash flow into immediate certain cash, at a substantial discount to face value. It removes servicing cost and management attention. The trade-offs are price, and loss of control over how the purchaser treats former clients — a live reputational exposure in markets where the institution continues to lend.
The decision test
For any account or cohort, pursuit is justified where:
Expected recovery × probability of success, discounted, exceeds the cost of pursuit
Most portfolios contain a large tail of small written-off balances where this test fails on every route. Recognising that early — and stopping — is a legitimate management decision, not a failure. Continuing to work accounts with negative expected value is a real and common cost.
Accounting Treatment
Recoveries on previously written-off loans are recognised in profit or loss when received. Presentation varies: commonly as a reduction of the impairment charge, sometimes as other operating income. Whichever is used should be applied consistently and disclosed.
Written-off accounts are not restored to the gross loan portfolio when partial recovery occurs. They remain derecognised, with the recovery flowing through the income statement.
Institutions should maintain memorandum records of written-off accounts — for continued pursuit, credit bureau reporting, and to prevent a previously written-off borrower being re-originated without the history surfacing. This last control is more frequently missing than it should be.
Under IFRS 9, expected recoveries also feed the ECL calculation before write-off, through the LGD parameter. Recovery data therefore has two distinct uses: cash generation after the fact, and loss estimation beforehand.
Conduct and Legal Constraints
Recovery activity is subject to the same consumer protection standards as collections, and often to greater scrutiny because the accounts are older and the tactics more aggressive.
- Prohibited practices — harassment, threats, contact at unreasonable hours or frequency, disclosure of the debt to employers, neighbours or community contacts, public shaming, seizure without legal process, and misrepresenting the borrower's legal position or the consequences of non-payment.
- Statute of limitations / prescription. Debts become legally unenforceable after a defined period, which varies by jurisdiction and may be reset by acknowledgement or part payment. Pursuing or threatening action on a time-barred debt is restricted in many jurisdictions.
- Settlement documentation. Any negotiated settlement should be documented, with a clear written statement of what remains owed, if anything, once paid.
- Data protection. Retention, sharing with agencies or purchasers, and credit bureau reporting all require a lawful basis.
- Agent oversight. Contractual conduct standards, audit rights and complaint monitoring, since liability and reputation stay with the institution.
Requirements vary by jurisdiction; confirm applicable rules with your regulator.
Key Recovery Metrics
MetricDefinitionNominal recovery rateRecoveries as a percentage of amounts written offEconomic recovery rateRecoveries net of collection costs, discounted to present valueCohort recovery curveCumulative recovery on a write-off vintage, by months elapsedLoss given default (LGD)1 − economic recovery rate; input to ECLCost to recoverRecovery costs as a percentage of amounts recoveredTime to recoveryMedian or weighted-average months from write-off to cash receivedLoan loss rate(Write-offs − recoveries) / average gross loan portfolioSettlement discount rateAverage discount accepted on negotiated settlements
Frequently Asked Questions
What does recovery mean in lending? Cash collected on a loan after default, and most commonly after the loan has been written off and removed from the balance sheet.
What is the difference between collections and recovery? Collections works delinquent accounts still on the books, aiming to return them to current. Recovery works defaulted or written-off accounts, aiming purely at cash at acceptable cost.
How is recovery rate calculated? The reliable method is cohort-based: cumulative recoveries on a defined write-off cohort divided by that cohort's original written-off balance. The simpler period ratio — recoveries divided by write-offs in the same period — is distorted by the timing of write-offs.
Does a write-off mean the lender stops trying to collect? No. Write-off is an accounting derecognition, not a legal waiver. Unless the debt is formally forgiven, recovery efforts may continue subject to conduct rules and any limitation period.
How does recovery relate to LGD? Loss given default is one minus the recovery rate, measured net of recovery costs and discounted for time. Observed recovery experience is what should ground the LGD assumption in an expected credit loss model.
When does most recovery happen? Early. A large majority of what will ever be recovered typically arrives within the first twelve months after write-off, with a long, thin tail thereafter — which is why speed of placement matters more than persistence.
Is it worth pursuing every written-off loan? No. Pursuit is justified only where expected recovery, discounted and weighted by probability of success, exceeds the cost of pursuing. Most portfolios have a tail of small balances where no route passes that test.