QuickPaisa: LSP-sourced personal loans and the limits of 'cosmetic' compliance
Sanchi Finance is an NBFC-ICC with ₹1,850 crore AUM and 31 branches. It plans 'QuickPaisa', a 90-day personal loan of ₹8,000–₹50,000 sourced entirely through a fintech LSP app with 2.4 million installs. Under the proposed structure the LSP identifies borrowers and runs its own scorecard; Sanchi remains lender of record. Disbursal is to happen from an LSP-held nodal account, net of a ₹299 platform fee and a 1.5% processing charge retained by the LSP. No Key Fact Statement is proposed; the LSP's terms run to 14 pages on its website. The app seeks contacts, photo gallery and SMS permissions at onboarding, and the LSP's recovery team calls defaulters' contacts and employers. Sanchi proposes a 7% first-loss default guarantee from the LSP, adjusted against monthly settlement. The compliance head calls these 'cosmetic fixes' and wants your sign-off so the product can go live in 15 days.
1. Identify each material breach of RBI's digital lending framework in the proposed structure and state the compliant alternative for each.
Read the worked answer
1. Funds flow (most serious). Disbursal and repayment must move directly between the borrower's bank account and the regulated entity's (RE's) bank account, with no pass-through or pool account of the LSP. An LSP-held nodal account is a structural breach, not a documentation issue. Fix: Sanchi's own account disburses; the LSP is paid by Sanchi, never from borrower funds.
2. LSP charging the borrower. The ₹299 platform fee and 1.5% processing charge deducted at source are, in substance, charges levied by the LSP on the borrower. All charges must be borne by the RE and disclosed in the Key Fact Statement (KFS).
3. Missing KFS. A KFS in English and the language of the borrower is mandatory and must carry the all-in cost/APR, all fees, the LSP's name, the grievance redressal officer, the cooling-off period (at least 3 days for loans above 7 days' tenure) and the recovery mechanism. Absence of a KFS alone would fail an inspection.
4. Data permissions. Contacts, gallery and SMS access are disproportionate. Only minimal, need-based data may be collected, with explicit, purpose-limited consent and no access to contact lists or media.
5. Recovery conduct. Calling the borrower's contacts and employer is prohibited; recovery must be by trained agents under a board-approved policy, restricted to the borrower and guarantor.
6. FLDG. 7% exceeds the 5% cap on the loan portfolio.
2. Sanchi argues the 7% 'first-loss default guarantee' is really a marketing support fee and should not be counted against the FLDG cap. Evaluate this position and explain how a compliant guarantee arrangement must be documented and operated.
Read the worked answer
The argument fails on substance. The FLDG clarification treats as FLDG any arrangement where an LSP effectively absorbs first loss on the RE's portfolio, whatever it is labelled - marketing support, sourcing fee, portfolio buy-back or settlement adjustment. Regulators look at risk allocation, not nomenclature: if Sanchi's loss falls by 7% before the LSP's own money is exhausted, it is a default guarantee.
Compliant design:
- Cap the arrangement at 5% of the loan portfolio, computed on the portfolio, not per loan.
- Recognise the cover only in permitted forms: cash deposit with the RE, a fixed deposit lien, or a bank guarantee in the RE's favour. A mere contractual promise is weak cover.
- Set out invocation mechanics in the contract: the RE should invoke within a defined window (commonly 120 days from default) and the proceeds are credited to the RE.
- Keep the credit decision and credit risk with Sanchi. The LSP's scorecard can be an input, but the RE must decide and must not be contractually obliged to accept every sourced file.
- Disclose the arrangement to the board and in the RE's disclosures.
A 'marketing fee' that is adjusted against defaults is a guarantee in disguise; supervisory action would follow, and auditors would be entitled to recharacterise it. Recasting it does not remove the 5% ceiling - it converts a correctable design flaw into a misrepresentation.
3. You are asked to sign off for a 15-day go-live. What is your recommendation, and what minimum sequence would you require before launch?
Read the worked answer
Recommendation: do not sign off. Four of the issues - funds flow, KFS, data permissions and FLDG quantum - require product, contract and integration changes, not text edits. A launch on this design would create a live, documented breach for every loan written, and remedial migration of a live book is harder than fixing it now.
Minimum sequence (indicative 8–10 weeks, run in parallel):
- Week 1–2: Board-approved digital lending policy and recovery policy; renegotiate the LSP agreement - direct disbursal, no borrower-level LSP fee, FLDG at 5% with a permitted instrument, credit decision retained by Sanchi.
- Week 2–4: Rebuild the app permission set and consent screens; data deletion on consent withdrawal; version the KFS generator in two languages.
- Week 3–5: Reconfigure the loan management system so disbursal, servicing and repayment are Sanchi-controlled; test the cooling-off and foreclosure path.
- Week 5–6: Internal audit and legal review of the flow, KFS, grievance routing and LSP disclosure in the app.
- Week 6–8: Pilot with a capped cohort; monitor complaint ratios, TAT and recovery call recordings.
Also note the RE retains accountability for LSP conduct - 'the LSP did it' is not a defence. Track LSP concentration, keep an exit plan, and put a named grievance officer with ombudsman details inside the app.