#2: How I’m Thinking About Starting an NBFC (After Looking at Too Many Portfolios)

Somewhere between the tenth and twentieth portfolio I reviewed, I realised I was no longer surprised by failures.

The products kept changing. Home loans, MSME, EV, solar, supply chain. Secured, unsecured. But the reasons things went wrong started repeating themselves. Sometimes slowly, sometimes all at once.

Analysing portfolios is what I do for a living. Big ones, small ones, clean ones, and the kind people stop talking about in meetings. Over time, this work quietly changes how you think. Not in dramatic ways, but in ways that make you uncomfortable with simple answers.

At some point, a question began to form in the background.

If I ever got the chance, how would I start an NBFC?

Not in theory. Not with perfect conditions. But with the kind of constraints most people have.

Starting with my own limitations

If I’m lucky, I may be able to raise modest capital to begin with.

In that world, my credit rating is unlikely to be anything better than BBB- at best. That implies a blended cost of funds north of 12–13 percent.

Once you accept this number, many popular lending ideas quietly stop making sense.

At this cost of funds, I don’t have the luxury of being loose on operating expenses (opex) or credit costs. Both need to be tightly controlled, and more importantly, predictable. Volatility is expensive when money itself is expensive.

High opex can still be worked down with discipline and effort. Volatile credit costs are far harder to fix. Especially when borrowers sit at the margin.

Low-ticket, marginal borrowers tend to be affected by even small economic blips. A delayed payment from a customer, a weak local season, a short-term slowdown. Things outside the lender’s control quickly show up in collections.

When cost of funds is high, you simply can’t afford parts of the business where you have limited control over outcomes.

What repeated portfolio reviews taught me to avoid

Early on, I used to admire fast-growing books. Large disbursement numbers feel reassuring. But after watching enough of them unwind, I’ve grown wary of growth without advantage.

Take two-wheeler loans. On paper, they look simple. In practice, a new NBFC with a 13 per cent cost of funds isn’t competing for prime borrowers. Those customers already have cheaper options.

What remains are rejected leads and marginal profiles, often in tier-3 towns and beyond. Add low ticket sizes, field-heavy operations, and collections friction, and the model starts relying on one fragile assumption: that behaviour stays benign.

Hope shows up frequently in post-mortems.

I’ve also noticed that spreading across many states early often feels like diversification but behaves like cost inflation. Each new geography adds complexity long before it adds resilience.

Low-ticket lending, even when secured, taught me a similar lesson. Growth looks fast, but opex grows faster. Credit costs don’t just rise, they swing. And when borrower resilience is low, the lender absorbs every small shock.

Over time, these patterns made me more cautious, not more confident.

Narrowing the box, reluctantly

Given these constraints, my thinking has gradually narrowed.

If I were to attempt this, I’d likely choose one of the following paths and stay disciplined about it:

• Higher-ticket loans, say ₹20 lakh and above, to more established MSMEs with visible cash flows.

or

• Fresh salaried borrowers who are new to credit, acquired with tight underwriting.

Not both. One segment, executed well, at a scale that justifies profitability.

Both can typically be priced in the 15–17 per cent range. Lending meaningfully beyond this often invites volatile credit costs, which eventually reflect into higher collection expenses, and consequently in opex.

This isn’t about taking less risk. It’s about taking risk you can observe, influence, and manage.

In smaller books, underwriting discipline matters more than clever structures. Cash flow understanding, bureau behavior, field checks, and simple curiosity about how a borrower operates start carrying more weight than templates.

I’ve also grown uncomfortable with models that only work at 18–20 percent pricing and beyond. I’ve seen too many portfolios where an extra 1–2 percent yield was quietly wiped out by collections inefficiency alone.

If a model cannot survive at 15–17 percent for most borrowers, it’s probably not as robust as it appears.

Beyond 20 percent pricing, in my experience, you’re entering a microfinance-like zone where group structures and social collateral matters. Outside those frameworks, making money consistently becomes difficult. Now I am hearing that JLG/SHG model is broken for good.

How my view on scale has changed

Earlier, I believed scale solved most problems. Now I’m less sure.

Expansion, I’ve realized, often arrives before profitability and then claims it was necessary all along. Geographic spread improves optics, especially with rating agencies, but it also increases distance between decision-makers and ground reality.

If I were building this from scratch, I’d rather grow slowly, backed by profitability and fresh equity, not pressure to impress.

Branches would exist, but technology would handle the unglamorous work. This wouldn’t be a fully digital story, but it also wouldn’t be a paperwork-heavy institution stuck in another decade.

I’d also resist the temptation to launch multiple products early. Until net worth crosses something meaningful, survival matters more than clever diversification.

What worries me about my own thinking

There are two things I remain cautious about.

The first is execution. Strategy always reads well on paper. Reality has a way of bending it. Processes drift. People interpret rules differently. Early success can make discipline feel optional.

I’ve seen well-thought-out credit philosophies diluted not by bad intent, but by speed.

The second is capital appetite. This approach is slow and deliberately boring. It may not excite venture capital or private equity, especially in a world trained to look for rapid scale.

But lending, I’ve come to believe, is not meant to be thrilling. It’s meant to be repeatable, resilient, and slightly dull.

Whether that trade-off is always worth it, I don’t fully know.

This is not a plan, just a snapshot

This isn’t a blueprint for starting an NBFC or I’d rather say this isn’t a blueprint for designing a loan portfolio,

It’s simply a snapshot of how my thinking looks today, shaped by what I’ve seen fail more often than succeed. I expect parts of this to change as I see more cycles and more outcomes.

If I ever do build something like this, reality will almost certainly disagree with at least half of what I’ve written here. The real question is which half, and how early I recognise it.

I’d genuinely like to hear how others think about this, especially those who’ve built lending businesses through different phases.

Because in lending, the most expensive lessons are usually the ones everyone thought they already understood.