Four numbers describe a loan book well enough to run one: outstanding balance, portfolio at risk, collection rate and write-off ratio. Each has a precise definition, each is routinely computed wrongly, and each can be made to look better than the book really is. Here are all three of those things, for each of the four.
Outstanding loan balance (OLB)
The total principal still owed on loans that have not been written off. Not the original amount disbursed, not principal plus interest, and not including loans you have already written off.
OLB is the denominator of nearly everything else, which is why getting it wrong quietly corrupts every other figure on the page. Two common errors:
- Including accrued interest. It inflates OLB, which makes PAR look smaller. If you do include it, include it in the numerator too, and say so.
- Including written-off loans. Also inflates OLB, also flatters PAR, and is the single most effective way to make a bad book look acceptable. Written-off loans leave the portfolio; they are tracked separately as recoveries.
Portfolio at risk (PAR)
PAR asks: what proportion of the money still out there is attached to a borrower who is behind?
Two details do all the work, and both are where PAR gets fudged.
The numerator is the whole loan, not the missed instalment. A borrower with KES 80,000 outstanding who is 31 days late on a KES 5,000 instalment contributes KES 80,000 to PAR 30, not KES 5,000. This is the point of the measure: the instalment is not the thing at risk, the relationship is. A "PAR" computed on arrears amounts is a different, much smaller and much less useful number, and it is startling how often it is presented as PAR.
The 30 is a parameter, and you must state it. PAR 1, PAR 30, PAR 60, PAR 90 are all legitimate and all different. PAR 30 is the common reporting standard in microfinance. A report labelled just "PAR" is ambiguous by construction.
A worked example
| Loan | Outstanding principal | Days past due | In PAR 30? |
|---|---|---|---|
| A | 120,000 | 0 | No |
| B | 80,000 | 12 | No — late, but under 30 |
| C | 45,000 | 31 | Yes, all 45,000 |
| D | 200,000 | 95 | Yes, all 200,000 |
| E | 55,000 | 0 | No |
OLB = 500,000. At risk over 30 days = 45,000 + 200,000 = 245,000. PAR 30 = 49% — even though only two of five loans are affected and only a few thousand shillings of instalments were actually missed. That is the measure working as intended.
The four ways PAR gets flattered
- Write off the bad ones just before reporting. They leave the numerator and the denominator, and PAR collapses. This is why PAR is never read without the write-off ratio beside it.
- Reschedule instead of collect. A restructured loan restarts its clock. Restructuring a genuinely distressed borrower can be good practice; using it to reset the days-past-due counter is not. Report restructured loans separately.
- Grow the book fast. New loans are never in arrears, so a rapidly growing denominator suppresses PAR while the underlying quality is unchanged or worse. Look at PAR by disbursement cohort, not just in total.
- Count arrears rather than balances, as above.
Collection rate
What proportion of what fell due in a period was actually collected in that period.
The trap is the numerator. If you include prepayments and arrears recovered from earlier periods, a collection rate can exceed 100% and stop meaning anything. Decide whether you are measuring on-time collection of current dues (exclude them) or total cash in against total due (include them, and expect it to swing).
Collection rate is the leading indicator that PAR is a lagging one. It moves within a month; PAR 30 needs thirty days to notice anything. If you can only look at one number weekly, look at this.
Write-off ratio
Its whole job is to be read next to PAR. A book with PAR 30 of 4% and a write-off ratio of 12% is not a good book; it is a book where the bad loans have been removed from view. A book with PAR 30 of 14% and a write-off ratio near zero is a book that is not facing its losses. Neither number is interpretable alone.
Note also that write-off is an accounting act, not a legal one. Writing a loan off does not extinguish the debt and does not stop recovery efforts, and recoveries on written-off loans should be tracked as income rather than quietly reducing the write-off figure.
How to report all four together
One line, monthly, per branch and consolidated, with the same definitions every month:
| Month | OLB | PAR 30 | Collection rate | Write-off ratio | Restructured |
|---|---|---|---|---|---|
| June | 14.2M | 8.1% | 94.0% | 0.4% | 2 loans |
| July | 15.8M | 7.4% | 91.2% | 0.3% | 9 loans |
| August | 17.9M | 6.9% | 88.5% | 0.3% | 14 loans |
Read that table the way a credit committee should: PAR is improving every month and it means nothing, because the book is growing 12% a month, the collection rate is falling five points, and the restructuring count has gone up seven-fold. PAR will catch up in about sixty days. This is exactly why the four numbers are reported together and why no single one of them is a KPI on its own.
Next: twelve things that cut PAR 30, in order of how little they cost you.