Loan Book Acquisition Due Diligence: How to Test a Pool
Loan Book Acquisition Due Diligence: How Lenders Evaluate a Portfolio Before Buying It
Last reviewed: November 2026
Loan book acquisition due diligence is the buyer's independent test of a loan portfolio before purchase: do the loans exist, match the seller's records, meet your credit standards, and perform as priced? Done well, it turns a seller's data tape into a price you can defend to your board and auditors.
This guide is for CEOs, CFOs and Chief Risk Officers at NBFCs, banks and housing finance companies buying pools through direct assignment. It covers the RBI's buyer obligations, the workstreams, the red flags and illustrative pricing arithmetic.
Key Takeaways
- Buyer rules sit in the RBI (NBFC – Transfer and Distribution of Credit Risk) Directions, 2025, issued on 28 November 2025 and updated as on 1 July 2026.
- Due diligence cannot be outsourced, and a rating cannot replace it.
- Portfolio-level checks are allowed only after loan-level checks on one-third of the pool by value and number, with the seller keeping 10%.
- The most expensive findings are usually data mismatches: tape balances that differ from the ledger, missing documents, or inconsistent overdue logic.
What Do the RBI Rules Require From a Buyer?
The RBI treats a bought loan like one you originated. You must apply the same standards, using your own people, under a board-approved policy.
| Requirement | What It Means in Practice |
|---|---|
| Board-approved transfer policy | Due diligence standards, pricing and risk limits written down before deals start |
| No outsourcing of due diligence | Your credit team tests the loans, even if advisers help |
| Loan-level checks by default | Portfolio-level checks only after one-third by value and number is tested loan by loan |
| Ratings are optional | A pool rating cannot replace your own testing |
| Ongoing monitoring | Stress tests and sensitivity analysis on acquired loans after purchase |
| Legal ownership transfers | Assignment documents and obligor consents where required |
| No credit enhancement | The seller cannot guarantee performance or buy loans back, except for warranty breaches |
The same directions set the seller's minimum holding period. Loans with tenor up to two years need three months on the seller's books, and longer loans need six months.
Stressed loans follow a separate chapter. They move only by assignment or novation, for cash, and an aggregate lender exposure of ₹100 crore or more triggers a Swiss Challenge auction.
Where the seller's exposure to a stressed borrower is ₹100 crore or more, it needs two external valuation reports. A buyer of stressed loans must also hold them for at least six months before selling them on, so build that lock-in into your exit assumptions.
Which Workstreams Make Up Portfolio Due Diligence?
Five workstreams cover most risks. Each needs a named owner, a test and a pass or fail threshold agreed before the data room opens.
| Workstream | What You Test | Red Flag |
|---|---|---|
| Data integrity | Tape totals against the seller's general ledger and bank statements | Unexplained balance gaps |
| Credit quality | Bureau data, income checks, policy exceptions, repayment history | Exceptions clustered in one channel or branch |
| Documentation | KYC, signed agreements, Key Fact Statements, mandates, security registration | Missing or unsigned files in the sample |
| Performance history | Static pools, roll rates, bounce rates, collection efficiency | Recent cohorts performing worse than older ones |
| Servicing and legal | Who collects after sale, cash flow routing, consent and assignment terms | Collections landing in an unrelated account |
The AI agents in due diligence overview explains where automation can speed document review. Your credit team must still own the conclusions.
How Should a Buyer Work Through the Data Room?
Start with the numbers, then move to files, then to behaviour. Each step narrows what the next step must check.
- Reconcile the tape. Match principal outstanding to the seller's ledger and recent collections to its bank statements.
- Re-run eligibility. Apply your own filters for holding period, classification, restructuring and residual tenor.
- Draw the loan-level sample. Cover at least one-third by value and by count, weighted towards higher-risk segments.
- Review files. Check identity, agreements, disclosures, mandates and collateral evidence for every sampled loan.
- Rebuild performance. Recompute days past due from repayment history rather than accepting the seller's field.
- Stress the cash flows. Model higher losses and faster prepayments, then price the pool on the stressed case.
A credit portfolio stress testing AI agent can run loss and prepayment scenarios on the tape. A portfolio concentration monitoring AI agent shows how the pool changes your exposure by geography, sector or partner.
Could your team reconcile a seller's tape to its ledger in days rather than weeks?
Talk to Our Lending Technology Specialists
Visit Digiqt to see how we build loan data pipelines and AI agents for lenders.
Illustrative Example: Testing a ₹250 Crore MSME Pool
This illustrative scenario uses round numbers. An NBFC is offered 12,000 unsecured MSME loans with ₹250 crore principal outstanding.
Step 1: sample size. To use portfolio-level checks on the rest, the buyer tests at least one-third loan by loan: ₹83.3 crore by value and 4,000 loans by count.
Step 2: findings. The tape shows ₹250 crore, but the seller's ledger shows ₹248.5 crore, leaving ₹1.5 crore to explain before pricing.
File review finds 300 loans worth ₹6 crore with missing agreements or KYC gaps. The buyer excludes them, leaving a ₹244 crore pool.
Step 3: retention and size. Because portfolio-level checks are used, the seller keeps 10%. The buyer acquires 90% of ₹244 crore, or ₹219.6 crore.
Step 4: expected loss. Assume an illustrative lifetime default rate of 4% and loss given default of 70%. Expected loss is 2.8%, or about ₹6.15 crore on ₹219.6 crore.
| Item | Amount (₹ crore) |
|---|---|
| Pool offered | 250.0 |
| Less: loans failing file review | (6.0) |
| Pool after exclusions | 244.0 |
| Buyer's 90% share | 219.6 |
| Illustrative expected loss at 2.8% | 6.15 |
The buyer now knows the yield spread must absorb about ₹6.15 crore of expected loss, plus servicing cost, before the deal adds value. If stress testing doubles losses, the price must reflect that case too.
Suppose the buyer instead pays a ₹2 crore premium over principal for a high-yield pool. The directions carry acquired performing loans at cost and require that premium to be amortised, so prepayments faster than modelled bring forward the premium write-off and cut the return.
What Should the Assignment Agreement Cover?
The agreement turns due diligence findings into enforceable terms. It must also stay inside the RBI's limits on what a seller may promise.
- Representations and warranties. Cover loan existence, documentation, KYC, eligibility and the accuracy of tape fields you relied on.
- Remedies for breach. The directions allow replacement within 30 days on original terms, or damages limited to direct losses.
- No hidden support. The seller cannot offer credit enhancement, liquidity support or buy-backs outside warranty breaches.
- Servicing terms. Set reporting frequency, remittance timelines, data formats, audit rights and a servicer exit plan.
- Data delivery. Require loan-level monthly files in an agreed layout, so performance can be loaded without manual rework.
Replacements and damages paid for warranty breaches must be reported to the RBI's Department of Supervision. Keep a clear audit trail linking each claim to the breached warranty and the affected loans.
Which Mistakes Do Loan Book Buyers Make?
These mistakes appear in otherwise careful deals because time pressure pushes teams to trust seller outputs.
- Accepting the seller's overdue field. Recompute days past due from raw repayment history.
- Sampling only the easy loans. A sample skewed towards seasoned, low-risk loans hides the real risk.
- Ignoring servicing after sale. If the seller keeps collecting, test its cash routing, reporting and exit terms.
- Pricing on the base case. Price on a stressed case, then treat the base case as upside.
What Happens After the Purchase?
Monitoring starts on day one. The directions expect buyers to watch acquired loans continuously, including stress tests and sensitivity analysis.
Load acquired loans into your own loan management system with a pool identifier, so performance can be compared with your originated book. An early delinquency warning AI agent can flag stress in the acquired pool before it shows in monthly reports.
Provisioning needs the same care. An expected credit loss intelligence AI agent can help keep loss estimates on acquired pools current as data arrives.
What Is Changing Next?
The directions require each loan transfer to be reported to a trade reporting platform once the RBI notifies it. Until then, NBFCs must keep a transaction database with MIS.
That makes clean, structured deal records a regulatory need, not just good practice. Buyers who capture pool, sample and exception data in systems will find the shift easier.
Want a repeatable due diligence process for every pool you buy?
Talk to Our Lending Technology Specialists
Visit Digiqt to see how we build lending platforms that onboard and monitor acquired loan books.
