Margin Compression at Deccan Meridian Urban Co-operative Bank: Gold Loans or Fintech Co-Lending?
Deccan Meridian Urban Co-operative Bank has ₹8,200 crore of advances, 41 branches across Maharashtra and Karnataka and a 210,000-member base. Over six quarters its net interest margin slipped from 2.94% to 2.31% as large private banks priced term loans below its own cost of funds. Fifty-eight per cent of the book is loan against property (LAP) to trading communities, CASA is 26%, and 71% of customers hold only one product. The board has two proposals. Option A: launch gold loans and micro-LAP in four trading clusters, at an estimated yield of 11.8% and credit cost of 0.9%. Option B: a co-lending tie-up with a fintech for unsecured personal loans at a 14.5% borrower rate, in which the partner sources and services and the bank funds 80% of every loan. The general manager's note states the partner will "handle" underwriting and collections. The chairman wants a decision in three weeks; both the RBI co-lending framework and the bank's by-laws, which restrict lending to members, constrain the choice.
1. Using Porter's five forces and VRIO, assess which option builds a defensible advantage rather than merely a higher yield. Show your reasoning chain.
Read the worked answer
Framing. Both options answer the same problem - undifferentiated LAP lending - so the test is not which yields more but which builds an advantage the bank can hold.
Five forces, applied narrowly.
- Rivalry: in LAP, price is the only visible variable. Against banks with 26%+ CASA and a ~3.1% cost of funds, a 2.31% NIM co-operative loses a price war by default.
- Buyer power: SME traders multi-source and switch on 25 bps; there is no switching cost.
- Substitutes: gold loans are a substitute for LAP for the same liquidity need. That makes Option A a repositioning within a known need, not a new market.
- Threat of entry: digital lenders reach these traders through UPI flows and the account aggregator; the branch relationship is a shrinking moat.
- Supplier power: depositors, not borrowers, are the binding constraint at 26% CASA.
VRIO.
- LAP relationships: valuable, moderately rare in these clusters, but imitable → competitive parity at best.
- Gold loans: valuable and rare within this peer set (many co-operatives lack vault, insurance and trained appraisers); imitable, but with a lag of roughly 18–24 months → temporary advantage.
- Unsecured co-lending: valuable, not rare, and the bank is not organised to exploit it - no unsecured underwriting, no collections, no behavioural data. VRIO fails at O, which predicts below-par returns, not parity.
Conclusion. Option A builds capability that can be extended; Option B rents someone else's and books their tail risk.
2. Quantify the two options roughly and recommend one, addressing the co-lending compliance issue and the by-law restriction.
Read the worked answer
Arithmetic (illustrative, ₹600 crore redeployed).
- Current LAP: 9.6% yield − 0.6% credit cost − 1.4% opex ≈ 7.6% net → ₹45.6 crore.
- Option A: 11.8% − 0.9% − 2.2% (vault, insurance, appraiser, security) ≈ 8.7% net → ₹52.2 crore. First-year delta is only ₹6.6 crore; the real case is capability plus cross-sell into a 71% single-product base.
- Option B: the borrower pays 14.5%, the partner takes roughly 4% as sourcing/servicing fee, leaving the bank about 10.5% before loss. The bank holds 80% of an unsecured loss curve that typically runs 5–8% in the first cycle, so net contribution can be negative - and the bank owns no customer, no data and no collection capability.
Compliance. RBI's co-lending framework requires the bank to retain at least 20% of each loan and to conduct its own credit appraisal; a note saying the partner will "handle" underwriting is not compliant in substance even if the agreement is signed. The 80:20 structure also concentrates adverse selection with the bank if the partner's incentive is volume.
By-laws. Membership-restricted lending caps the addressable base; gold and micro-LAP against member traders fits within it, unsecured retail sourced by a third party may not.
Recommendation. Option A, phased across four clusters, with Option B revisited only after the bank has its own unsecured underwriting and collections capability.
3. The chairman asks what would make you reverse the decision within twelve months. Set out the pre-mortem and the monitoring triggers.
Read the worked answer
Pre-mortem (why Option A fails by month 12).
- Capability gap: appraisers are hired but not certified; valuation variance between branches exceeds 10% and auditors qualify the book.
- Fraud: gold purity and ownership verification are weak in cluster branches with high cash handling; a single ₹3–4 crore fraud freezes the product.
- Pricing drift: clusters get crowded, yields slide to 10.2%, and the 8.7% net collapses to LAP levels with worse opex.
- Funding: gold loans are short-duration; if 26% CASA does not improve, the bank funds them with expensive term deposits and the margin gain evaporates.
- Opportunity cost: four clusters absorb branch management attention and micro-LAP credit costs overshoot at 1.8%.
Monitoring triggers (monthly to the board risk committee).
- First-60-day delinquency on micro-LAP above 1.5% → halt disbursement in that cluster.
- LTV breaches (RBI caps gold loan LTV at 75%) on more than 2% of tickets, or any unreconciled ornament.
- Yield below 10.8% for two consecutive quarters → reprice or exit the cluster.
- Appraiser certification coverage below 90%, or valuation variance above 8%.
- Cost of funds above 6.4% → treat as a funding failure, not a product failure.
The judgement. Reverse if two or more triggers fire together - that pattern signals the capability thesis is wrong, not merely that execution is slow. Acting on a single trigger risks abandoning a sound strategy on normal variance.