Delinquency

Delinquency is the state of a loan on which a scheduled payment has not been made by its due date. It begins the day after a missed instalment and continues until the arrears are cleared, the loan is restructured, or it is written off.

Unlike default, delinquency is a continuum rather than a threshold. A loan one day late and a loan two hundred days late are both delinquent; they are entirely different risks. This is why delinquency is always measured with an aging dimension attached, never as a single binary flag.

Delinquency is the earliest quantitative signal a lender has. Every downstream credit metric — default, non-performing classification, expected credit loss staging, write-off — is a consequence of delinquency that was not resolved.

How Days Past Due Is Calculated

Days past due (DPD) is the number of days elapsed since the oldest unpaid contractual instalment fell due.

This is the point most often got wrong. DPD is not measured from the most recent missed payment, and it is not reset by a payment that fails to clear the oldest arrear. A borrower who pays something every month but never catches up has an ever-increasing DPD, not a perpetually low one.

DPD = Reporting date − Due date of the oldest instalment still unpaid

Where the calculation gets complicated

Payment allocation. Where a partial payment arrives, the allocation waterfall determines which instalment it clears — and therefore what the DPD becomes.

Example. Instalments fall due on 5 January, 5 February and 5 March. The borrower pays nothing in January, makes one instalment payment in February, and nothing in March. At 31 March:

• Oldest arrear first: Oldest unpaid instalment is 5 February → DPD at 31 March is 54 days
• Applied to current instalment: Oldest unpaid instalment is 5 January → DPD at 31 March is 85 days

Same cash, same borrower, a 31-day difference in reported delinquency and potentially a different delinquency bucket. Allocation logic must be defined in policy, configured consistently in the system, and understood before any DPD figure is interpreted.

Waterfall composition. Where the waterfall clears penalties and fees before principal and interest, a payment can be fully consumed by charges while the instalment itself remains unpaid — so cash is received and DPD does not move.

Grace periods. Contractual grace before a payment is treated as late. Policy should state whether DPD counts from the due date or from the end of grace, and apply it consistently.

Materiality thresholds. Many frameworks require arrears to exceed an absolute or relative threshold before the DPD clock starts, preventing trivial shortfalls from generating delinquency.

Non-business days. Whether due dates falling on weekends or holidays roll forward, and whether DPD counts calendar or business days.

Contractual vs Recency Delinquency

Two fundamentally different ways of expressing lateness:

Contractual delinquency measures against the original repayment schedule — how far behind the contract the borrower is. This is the standard basis for regulatory reporting, provisioning and credit risk measurement.

Recency delinquency measures days since the last payment of any size was received, regardless of whether it covered a full instalment.

Recency looks more forgiving and is sometimes used operationally to gauge borrower engagement — a borrower paying something is more likely to cure than one who has gone silent. But it systematically understates risk: a borrower paying a fraction of each instalment shows near-zero recency delinquency while falling steadily further behind on a contractual basis.

Use contractual delinquency for all risk, provisioning and reporting purposes. Recency is a collections triage input, not a credit metric.

Delinquency Buckets and Aging Analysis

Delinquent accounts are grouped into buckets by DPD, forming an aging analysis:

• 1–30 days (Early / Bucket 1): Automated reminders, first contact
• 31–60 days (Bucket 2): Intensified contact, field follow-up
• 61–90 days (Bucket 3): Formal escalation, workout discussion
• 91–180 days (Non-performing): Structured resolution, restructuring or enforcement
• 180+ days (Late / recovery): Legal, agency, write-off consideration

The aging distribution is more informative than any single delinquency figure. Two portfolios can share an identical overall delinquency rate while one is concentrated in the 1–30 bucket and the other in 90+ — the first is a collections timing issue, the second is a credit quality problem.

Roll rates — the proportion of balances moving from one bucket to the next in the following period — are the forward-looking read on the aging table, and the earliest reliable indicator of where a portfolio is heading.

How Delinquency Is Measured

Several distinct ratios all get called "the delinquency rate," and they give very different answers on identical data.

Worked example

A portfolio of 4,000 loans totalling 10,000,000. Of these, 340 loans have at least one payment overdue. Their combined outstanding balance is 1,200,000, and the actually overdue instalments within that total 220,000.

• Arrears rate: Overdue amounts / Gross loan portfolio = 220,000 / 10,000,000 = 2.2%
• Portfolio at risk (PAR>0): Full outstanding balance of delinquent loans / GLP = 1,200,000 / 10,000,000 = 12.0%
• Delinquency rate by count: Delinquent loans / Total loans = 340 / 4,000 = 8.5%

Three defensible figures spanning 2.2% to 12.0%, describing the same book.

The arrears rate is the misleading one. It counts only the instalments actually overdue and ignores the rest of the balance on those loans — as though the remaining principal on a defaulting borrower's loan were unaffected. It systematically understates exposure and should not be used as a headline credit quality measure.

PAR is the standard convention in microfinance and small-loan lending precisely because it counts the full outstanding balance of any loan with arrears. Always confirm which basis a quoted figure uses.

Count-based measures are useful operationally — they indicate workload — but they weight a small loan and a large one equally, so they are not a risk measure.

What Causes Delinquency

Separating the causes is the most useful diagnostic available, because the correct response differs entirely for each.

• Cannot pay — capacity: Income shock, business failure, illness, seasonal downturn, over-indebtedness across multiple lenders. Response: assess viability, restructure where there is genuine capacity to service a revised schedule.
• Will not pay — willingness: Capacity exists. Often a dispute over charges, dissatisfaction with the lender, perception that others are not paying, or a calculation that consequences are minimal. Response: escalation and consequence.
• Process failure: The borrower intended to pay and something broke — a failed debit mandate, insufficient balance at the debit attempt, payment made to the wrong account, agent unavailable, group meeting missed, payment received but misapplied. Response: fix the process.

The third category is consistently larger than institutions expect and the cheapest to resolve. A material share of "delinquency" in digitally collected portfolios is mandate and timing failure rather than borrower behaviour, and treating it as credit risk both wastes collections effort and damages relationships with borrowers who did nothing wrong.

Why Delinquency Matters

• It is the leading indicator. Delinquency moves weeks before default, months before non-performing classification, and longer still before write-off. It is the only credit metric that responds fast enough to act on.
• Cure probability decays with age. Early-bucket accounts cure at far higher rates and at far lower cost per account than aged ones. The economics strongly favour intervention in the first thirty days.
• It drives provisioning. Under IFRS 9, 30 days past due carries a rebuttable presumption of significant increase in credit risk, moving an exposure to stage 2 and replacing a 12-month allowance with a lifetime one.
• It costs money directly. Collections effort, staff time, field visits and system contact all scale with delinquent volume.
• It signals upstream problems. Rising early-bucket delinquency in recent origination cohorts is an underwriting signal, not a collections signal.

Common Pitfalls

• Measuring DPD from the wrong instalment — from the most recent miss rather than the oldest unpaid one. This understates aging, sometimes severely.
• Ignoring allocation effects. As shown above, the waterfall determines DPD. Two systems with different allocation logic will report different delinquency on identical cash flows.
• Quoting the arrears rate as the delinquency rate. It understates exposure several-fold against PAR.
• Re-ageing on restructuring. Rescheduling resets the DPD clock, so a portfolio can be cleaned of delinquency without any borrower position improving. Restructured volume must be reported alongside delinquency for either figure to mean anything.
• Blending products. Weekly-repayment group loans and monthly individual loans have structurally different delinquency profiles and should not share a single aging table.
• Ignoring seasonality. Agricultural cycles, school fee periods and festival seasons produce predictable delinquency peaks. Month-on-month comparison without a seasonal baseline generates false alarms in both directions; compare against the same period last year.
• Treating the blended figure as actionable. Portfolio-level delinquency tells you a problem exists. Only the cut — by vintage, branch, officer, product, sector and loan cycle — tells you where.