Repair the franchise, not just the NPA ratio
PAT 86 vs prior loss; ARC-adjusted ratio matters.
Watch: WATCH; verify new-vintage cash returns.
Home / Research / Fusion Finance Limited
Repair the franchise, not just the NPA ratio
PAT 86 vs prior loss; ARC-adjusted ratio matters.
Watch: WATCH; verify new-vintage cash returns.
Disbursement 1,726, up 33%.
Watch: Track actual net asset growth.
PPOP 89→125 YoY.
Watch: Separate operating income from recoveries.
−22 to 86; Q1 62.
Watch: Normalize legacy recoveries and tax.
SAGE PRO Institutional Equity Research | 16 November 2026 | NSE: FUSION | Q2/H1 FY2027, 30 September 2026
Decision: WATCH — turnaround evidence is encouraging, but require clean vintage profitability. Reference close ₹165.80 on 13 November 2026 (Sage PRO canonical price); horizon 12–24 months. Figures ₹ crore unless stated. Original issuer Q2 presentation and call with Q1 comparatives. Reported versus ARC-adjusted credit quality is explicitly separated. Available-source longform; full statutory equity/share-count, asset sale consideration and peer valuation remain open.
Key takeaways
The branch network and borrower history are valuable only if they improve repayment economics without encouraging multiple-lender debt.
The September quarter shows a material turnaround. AUM ₹8,007 crore (+14% YoY), disbursement ₹1,726 crore (+33% YoY), PPOP ₹125 crore (+40% YoY), and PAT ₹86 crore versus a ₹22 crore loss a year earlier.
Q1 PAT ₹62 crore. Reported GNPA 1.6% compared with 2.51% June; net NPA 0.11%. [A,B] Management reports approximately 4% annualized ROA and 13% ROE. This is not merely a reported ratio recovery: PPOP and funding also improved.
The differentiated thesis is that Fusion may earn a structurally better return by combining tighter borrower leverage controls with lower funding cost and branch rationalization. The remaining challenge is to separate true credit repair from changes in accounting exposure and collections.
Counter-thesis: disposal of delinquent assets, recovery of written-off loans, and a favorable funding window may temporarily elevate results. A return to faster AUM growth before new loan cohorts fully season could recreate credit stress. The stock deserves attention, not an unqualified turnaround multiple.
| Metric | Q2 FY26 | Q1 FY27 | Q2 FY27 |
|---|---|---|---|
| AUM | ~7,024* | — | 8,007 |
| Disbursements | ~1,298* | broadly similar | 1,726 |
| NIM (issuer metric) | 10.85% | 11.93%* | 12.52% |
| NII | — | — | 265 |
| PPOP | 89 | ~102* | 125 |
| PAT | -22 | 62 | 86 |
| Credit cost amount | — | 39 | 39 |
| GNPA | — | 2.51% | 1.6% |
| Marginal funding cost | 12.16% | 10.10% | 9.35% |
Key takeaways
Q2 PPOP increased ₹36 crore YoY and approximately ₹23 crore sequentially. PAT improved ₹108 crore YoY from a loss; the magnitude indicates both operating and below-PPOP changes. The PPOP less reported credit cost bridge is 125−39=86, numerically equal to PAT.
This is not a verified PBT-to-PAT bridge; tax, recoveries, other income and accounting treatment may be present. Do not infer zero tax from rounded amounts. Require statutory income statement and loan-sale accounting before building a full normalized earnings model.
PPOP growth is particularly useful because it reflects improvement before reported loan losses. NIM rose from 10.85% prior year to 12.52% and marginal funding cost fell 281bp YoY. At a given asset yield, lower funding cost can materially improve earnings, but the marginal rate is not the blended funding rate and future refinancing benefit depends on actual maturity.
This is stronger evidence than a net profit rebound alone. H2 performance must show recurring PPOP without the need for large one-time recoveries.
Key takeaways
Multiple-lender exposure was an important mechanism in microfinance stress: borrowers may continue paying one lender by borrowing from another, masking deterioration until access to refinancing stops. Fewer lender relationships reduce this risk, but do not prove income adequacy. Test household-level debt service, total indebtedness, borrower income verification, repeat loan purpose and cohort repayment behavior rather than count alone.
Same-day X-bucket collections 97.9%; monthly gross and net collection efficiency ~99.7% and 99.84% in the quarter. September collections recovered after rain/flood disruptions in five markets; October management reported ~99.78% gross.
These measures are current or specific collection definitions, not total arrears recovery. The strongest proof will be low flow from current into delinquent buckets at 3/6/9 months of comparable vintages.
Management's revised credit-assessment templates should be assessed against control cohorts, including approval rates and actual loss.
The core JLG franchise is complemented by premium JLG for borrowers graduating beyond income thresholds and individual/secured MSME lending. The Vishwas product represented about 20% of MFI September/October disbursement according to management.
Approximately 10% of MSME disbursements originated through the MFI branch network; MFI spokes increased from 50 to 100. These initiatives may reuse distribution and borrower information, but individual cash-flow underwriting and collateral documentation differ from group lending.
Graduation must reflect verified income and repayment history, not regulatory arbitrage or larger tickets to stressed households.
MFI AUM per branch improved from ₹5.1 crore to ₹5.4 crore (~5.9%). Management plans to rationalize ~102 branches and open ~50 in attractive markets. This can reduce fixed costs and improve portfolio density; however, closures may impair collections if service access deteriorates.
Compare existing and new branches on actual cash contribution after field costs, incremental delinquency and borrower retention. New LOS/LMS implemented in about 15% of branches, with full transition targeted by fiscal year end; system migration introduces operational risk, and realized benefits should be verified via processing quality and collections.
Key takeaways
[B, pp.2,5] This is a critical investment distinction: disposal can transfer stressed exposures out of the reported book, while actual borrower cures and new cohort quality determine future loss. The company sold secured loans, including ~₹3.5 crore previously written-off, to an ARC; the transaction's consideration, retained security receipts, guarantees and cash realization must be reconciled from statutory filings. Do not assume all economic risk has disappeared.
Provision stock ₹209 crore includes management overlay ₹19.5 crore; Stage 3 coverage 93.4%, Stage 2 coverage 67.3%. Reported quarterly credit cost ₹39 crore, flat sequentially, and management's medium-term objective 2–2.5% credit cost.
High provisioning is useful loss absorption, not a guaranteed release. Management also reported >₹30 crore recoveries from 90+ DPD exposures, including ₹21 crore from written-off accounts. Recoveries can add real cash value, but legacy recoveries are not necessarily recurring earnings.
Model new book expected loss separately from old recoveries, ARC transaction effects and provisioning changes.
Geographic stress remains a relevant monitoring item. In the October Q&A, management cited Maharashtra book ~₹117 crore with October collection ~98.84%, lower than prior ~99.4%, and Karnataka ~99.95%. These are management-reported, not independent district statistics.
Early stress can be geographically concentrated and may be masked by high group average collections. Monitor actual PAR flows, moratoria, borrower cash flow and post-flood cures. Do not infer all geographies are equally resilient.
Annual illustrative sensitivity: a 50bp incremental credit-loss rate on closing AUM ₹8,007 crore represents ~₹40 crore pre-tax, about 12% of Q2 annualized PAT ₹344 crore. A 100bp loss-rate increase ~₹80 crore. Actual average assets and retained risk differ. A lender in credit recovery should be valued on through-cycle loss cost, not on one quarter's benign collection flows.
Key takeaways
[B, pp.5–6] This is favorable because lower marginal funding cost can support net interest margin even without higher loan yields. But secured/securitized funding, guarantees and program terms can vary; a current low marginal rate is not guaranteed in a tightening cycle. Track actual blended cost, maturity and funding diversity.
Reported liquidity ₹1,661 crore, undrawn sanctions ₹2,800 crore and pipeline ₹4,500 crore. Do not add all three as cash. Undrawn sanctions are conditional access; pipeline is not committed liquidity. Capital adequacy 36.75% gives room to grow, but low leverage can suppress ROE.
At ~4% annualized ROA and 13% ROE, the next test is deployment into profitable seasoned assets without overleveraging. Management's stated medium-term AUM growth 20–25%, AUM ambition ₹10,000 crore, ROA 4%+ and credit cost 2–2.5% are targets, with timeline slightly calibrated due H1 disruption.
Neither capital adequacy nor a rating upgrade proves future access under stress.
Risk monitoring should include partner securitization retention, collection servicer obligations, borrower concentration and asset-liability mismatch. A low marginal funding rate is valuable only if asset repricing and realized losses allow sufficient net return. No assumption of future equity capital or dilution is embedded without statutory terms.
Key takeaways
It deliberately brackets Q2 annualized PAT ₹344 crore and allows downside if credit recovery or margin moderates.
| Case | AUM growth | Assumed PAT/avg AUM | Illustrative PAT ₹cr |
|---|---|---|---|
| Bear | 8% | 2.0% | 167 |
| Base | 20% | 3.5% | 308 |
| Bull | 25% | 4.2% | 378 |
A terminal P/B sensitivity: sustainable ROE 8%/14%/18%, cost of equity 16%/15%/14%, growth 3%/5%/6% gives ~0.38x/0.90x/1.50x. A longer recovery and growth phase could add value; these are not target prices. At ₹165.80 reference close, exact attributable equity, issued shares and current book must be independently reconciled before making an upside call. Investors should not apply the peak pre-crisis multiple to a newly repaired balance sheet.
Peers CreditAccess Grameen, Satin and Muthoot Microfin offer relevant credit-cycle and collection comparisons, but reporting definitions, overlays, capital raises, product mix and asset sales differ. Compare core current-portfolio vintage loss and cash-adjusted ROE rather than reported GNPA. A lender with lower reported GNPA because of a loan sale may not have the better underlying origination franchise.
| Metric | Current evidence | Decision test |
|---|---|---|
| GNPA | 1.6% reported / 2.28% ARC-adjusted | Track comparable ex-disposal ratio, cash proceeds and residual risk |
| New borrower leverage | 83% ≤2 lenders | Continue income/debt assessment and cohort loss |
| Credit cost | ₹39cr Q2; medium-term aim 2–2.5% | Normalize without relying on recovery or provision release |
| Funding | 9.35% marginal | Verify blended cost and refinancing conditions |
| Branch productivity | ₹5.4cr MFI AUM/branch | Net contribution after rationalization and collection cost |
| Earnings | ₹86cr Q2 vs ₹62cr Q1 | Recurring PPOP and cash-adjusted PAT |
| Capital | CRAR 36.75% | Deploy prudently; preserve liquidity |
Catalysts: Q3 confirmation of low net flows in the newly underwritten book, sustained funding benefit, improved branch productivity and clear ARC cash/retention disclosure. Thesis breakers: renewed borrower over-indebtedness, an unexplained difference between reported and economic NPA, loss acceleration as new cohorts season, or dependence on asset disposals and recovery income to sustain profit. Maintain watch. A turnaround becomes investable at a higher conviction level only when recurring earnings and actual book value support risk-adjusted valuation.
Analyst model
This is dated research, not a live market quote. Scenarios and historical performance are not guarantees.