Nepal's MSME financing gap is US$6 billion. The infrastructure that could close it is finally being built, and that is what makes P2P lending worth talking about.

Every few years a financial technology arrives in Nepal carrying more hope than it can hold. Peer-to-peer lending and crowdfunding have been carrying that hope for some time. The pitch is intuitive and, in a country like ours, genuinely moving: a small entrepreneur in Nepalgunj who cannot post land as collateral connects directly with savers willing to fund her, and the bank, its ledgers full of unsellable property, is simply routed around.

It is a good story. It has also, until now, been an incomplete one.

The Government has placed P2P lending and crowdfunding in the national budget. Monetary Policy for FY 2083/84 commits Nepal Rastra Bank to studying peer-to-peer transactions built on individual credit scoring. And in January 2026, the World Bank approved a US$95 million operation whose entire purpose is to build the credit infrastructure that Nepal has never had.

Those three things happened separately. Together, they change what is possible.

The size of the hole

Start with the gap, because it is larger than most people realise.

The World Bank's Sustainable and Inclusive Finance Project Appraisal Document, approved in January 2026, states plainly that "an International Finance Corporation (IFC) study estimates that the financing gap for MSMEs in Nepal is US$6 billion (around 14 percent of GDP)." The underlying estimate comes from the SME Finance Forum's MSME Finance Gap work of March 2025.

Fourteen percent of GDP. Not a market inefficiency to be tidied at the margins, but a structural absence.

And it sits beneath the most productive part of our economy. The same document records that MSMEs contribute more than 90 percent of Nepal's industrial GDP and around 70 percent of total national exports. This is not a peripheral sector waiting for charity. It is the sector that makes things and sells them abroad, operating without the credit it needs.

The exclusion is measurable, and the numbers are stark. Only 17 percent of MSMEs use banks to finance investments. Only 16 percent of start-up enterprises have access to capital from banks and financial institutions. Of formal SMEs, the SME Finance Forum finds just 39 percent unconstrained in their access to credit, with 17.7 percent partially constrained and 43.6 percent fully constrained. For women it is worse still: 52 percent of women-owned small businesses find it challenging to access finance, and only 20 percent of women borrow from a formal financial institution at all.

The World Bank's summary of why is worth reading closely, because it names three barriers: "inadequate collateral, lack of credit history, and limited financial literacy."

Hold those against Nepal's banking system and everything resolves.

The collateral trap

Roughly 74 percent of bank lending in Nepal is secured against fixed-asset collateral, according to the World Bank's own figures. That is not a preference. It is a substitute for information.

A bank that cannot assess whether a borrower will repay asks instead for something it can seize if they do not. Collateral is what lending looks like in the absence of a credit history. Which means the first two barriers the World Bank names are not two problems. They are one problem, seen from either end.

And the trap has now closed on the banks themselves. When Nepal's property market stopped moving, the collateral stopped being liquid. The World Bank notes that loan recovery "has been impacted by the decline in value of real estate, the primary collateral used to secure loans," and records non-performing loans climbing from 1.5 percent in 2019 to 4.6 percent by June 2025. Nepal Rastra Bank's own Macroeconomic Report puts gross NPLs at 5.6 percent by April 2026, with the watchlist category, the leading indicator of tomorrow's bad loans, rising from 6.7 percent in mid-2023 to 11.1 percent.

The collateral regime did not merely exclude the borrower without land. It eventually punished the bank that relied on it.

Now set that beside a second fact. Last year, private-sector credit grew 6.5 percent against a 12 percent target, in a banking system holding roughly Rs. 1.1 trillion in excess reserves, with lending rates down at 6.7 percent.

The money is not scarce. The bridge is missing.

The World Bank puts this more bluntly than I would dare to. In its lessons-learned section, it writes of "countries like Nepal where ample liquidity is available, however financial institutions do not lend to MSMEs due to perceived riskiness of such lending." Two institutions, working independently, have arrived at the same diagnosis. Liquidity is not the constraint. Information is.

What P2P is, and what it is not

Here is where enthusiasm usually outruns design.

A P2P platform does not create creditworthiness. It creates a venue. If the underlying information problem is unsolved, the platform does not solve it. It relocates it. Instead of one bank unable to assess a borrower, you have several thousand retail lenders unable to assess a borrower, with far less capacity to absorb the loss when they are wrong.

Notice that the World Bank's list of barriers did not stop at collateral and credit history. It also named limited financial literacy. That cuts in a direction the advocates rarely acknowledge. It is not only a constraint on borrowers. It is a constraint on the ordinary savers a P2P market would ask to become lenders, pricing credit risk on strangers, often for the first time in their lives.

An instrument meant to close a gap caused partly by low financial literacy cannot be built on the assumption that lenders will exercise sophisticated credit judgement. That is not a footnote. It goes to whether the thing is safe at all.

The international record is unambiguous. Jurisdictions that opened P2P markets before building the informational and supervisory scaffolding did not democratise credit. They produced platform failures, retail investor losses, and eventually a clampdown that set the sector back a decade. China is the cautionary extreme, but not the only case.

The lesson is not that P2P is dangerous. It is that P2P is a distribution innovation, and distribution innovations only work once the underlying credit-assessment problem is solved. Otherwise you have not disintermediated risk. You have redistributed it, toward the people least equipped to carry it.

Why the credit bureau is the whole story

If information is the binding constraint, then the credit reporting system is the thing that matters, and Nepal's is thin.

The World Bank's assessment is unsparing. The Credit Information Center Limited collects only negative information. Credit bureau coverage of the adult population, excluding microfinance borrowers, stands at around 10 percent, against 21 percent for South Asia. The bureau has no access to utility payments, insurance records or tax data. And there is no unique identifier linking data across sources, which leaves collection fragmented and inconsistent.

Read that and the 14-percent-of-GDP gap stops being mysterious. A financial system that records only who has failed to repay, and reaches only one adult in ten, cannot lend on character or cash flow. It can only lend against land. It is not choosing collateral. It has nothing else to choose.

This is why the credit bureau, not the platform, is the real subject of any serious P2P conversation. A credit score is not a technology. It is public infrastructure, like a land registry or a payments system. It converts a person's history into something a stranger can price. Once it exists and is trusted, a great many things become possible that were impossible before, and P2P lending is only one of them.

Without it, P2P is a marketplace in which nobody can read the price tags.

What changed in January

And this is the development that has gone largely unremarked.

The World Bank's US$95 million Sustainable and Inclusive Finance Project is not, despite the name, primarily a lending programme. Read the components. US$7.8 million goes to Component 2: Enhancing Credit Information Reporting Systems, and what it funds is precisely the missing architecture.

It will move CICL from negative-only reporting to a comprehensive regime capturing both positive and negative data. It will integrate non-traditional data sources, including mobile transactions and utility payments. It will develop credit scoring methodologies, including gender-sensitive models. And it will install a centralised Know Your Customer repository, so a customer verifies identity once and that verification is shared securely across institutions.

The targets are dated and specific. By June 2030: c-KYC adopted by all financial institutions; telco-score adoption in place; credit information collected from non-traditional sources; total borrowers with credit records rising from 7.8 million to nearly 10 million. Alongside it, US$87.2 million recapitalises the Deposit and Credit Guarantee Fund, whose credit risk reserve ratio stands at just 3.3 percent against 10 percent in India and 20 percent in Korea.

The stated purpose of all this, in the Bank's own words, is to deliver "lower collateral requirements, and improved reputational collateral for thin-file borrowers."

Reputational collateral. That phrase deserves to be lifted out and looked at. It is the entire answer to the collateral trap: the idea that a person's record of keeping their word can substitute for a plot of land. That is what a functioning credit bureau produces. And Nepal is now, for the first time, building one with money and a deadline attached.

The sequence is the policy

The consultative document Nepal Rastra Bank issued on P2P lending and crowdfunding is, read properly, not a document about platforms at all. It is a document about order of operations. Nearly every recommendation states what must exist before something else is permitted.

It proposes sequenced, pilot-first licensing: not an open market, but a sandbox or limited-license cohort of two or three operators, with caps on aggregate platform exposure, single-borrower exposure and single-investor exposure. Graduation to full licensing is conditional on demonstrated compliance, not elapsed time. Nobody earns a licence merely by surviving.

It then specifies prerequisite infrastructure, and this is where the argument closes. Full-phase P2P operation should not precede: an operational national ID linked to a unique financial identifier; a functioning credit information and scoring architecture with coverage extending beyond formally banked borrowers; centralised or interoperable e-KYC; and an accessible dispute resolution channel. Where these are incomplete, platform activity stays restricted in scale and product scope.

Now compare that list against what the World Bank is funding. Unique identifier: the Bank names its absence as a core deficiency, and the Credit Information Reporting Bill addresses it. Credit scoring architecture reaching beyond the banked: Component 2, explicitly including non-traditional data. Centralised e-KYC: funded, with a 2030 target. Dispute resolution: the Credit Information Reporting Bill carries dispute resolution and enforcement mechanisms, and the project tracks grievance redress at 100 percent.

The 2024 consultative document set four conditions. Three years later, an independent institution is financing all four, without being asked to.

That is not coincidence. It is convergence, and it is the strongest possible validation that the sequencing was right.

The document is equally clear about where losses must sit. Platforms are to be treated as intermediaries, not deposit-taking or risk-bearing entities. Any express or implied capital guarantee is prohibited. So is maturity transformation. So is commingling platform funds with lender funds; client money goes into escrow at a licensed institution. Platforms must disclose, in standardised form, historical default rates, recovery rates, fee structures, and the plain fact that principal is not guaranteed.

This is the ambiguity-killer, and it matters more than any other provision. Every P2P regime must eventually answer who bears the credit loss, and those that answer late answer badly. Ambiguity does not remain ambiguous. Under stress, it resolves into a bailout expectation.

It draws a perimeter around crowdfunding that Nepal will be grateful for later. Donation and reward-based crowdfunding may proceed under lighter-touch conduct rules. But equity and debt-based crowdfunding raise securities-law questions requiring coordinated treatment with SEBON, and a clear jurisdictional demarcation established before licensing begins. Conflating these instruments is how you write rules that fit none of them, and how you invite arbitrage between two regulators each assuming the other was watching.

And it insists that supervisory capacity be built in parallel with, not after, the licensing regime: a centralised repository receiving standardised loan-level data from every licensed platform, with capacity in data analytics, platform inspection and algorithmic audit. A regulator that licenses first and learns to supervise afterwards is not regulating. It is hoping.

The proposed sequence is explicit: consultation, legal and licensing framework, infrastructure readiness assessment, small-cohort pilot, evaluation against defined supervisory indicators, and only then a decision on full-phase operation. Each stage carries explicit exit criteria. Moving to the next without meeting them would expose the financial system to risks the current infrastructure cannot absorb.

That last sentence is the whole argument, and it is worth being blunt about what it implies. The pilot is permitted to fail. A pilot that cannot fail is not a pilot. It is a launch with a softer name.

What I would watch

Four questions will decide whether this arrives anywhere.

Does the scoring architecture actually reach the unbanked? This matters most and will be hardest. Bureau coverage stands at 10 percent of adults. A scoring regime that captures only those already inside the formal system reproduces the exact exclusion it was meant to cure. It would give us a beautifully scored view of the people banks already lend to, and remain blind to the US$6 billion. The World Bank was specific about who is missing: start-ups, thin-file borrowers, and the 80 percent of women who never approach a formal institution. Coverage, not sophistication, is the test.

Does the unique identifier arrive? Both the consultative document and the World Bank name its absence as fundamental. Without it, data from utilities, telcos and tax records cannot be reliably linked to a person, and the "non-traditional data" that is supposed to see the invisible borrower simply cannot be assembled. This is the keystone. Everything else waits on it.

Is the SEBON demarcation settled before licensing, or after? Before is a boundary. After is a dispute, conducted in public, with investors' money in the middle of it.

Do the exit criteria hold when they become politically inconvenient? They will be tested. A budget commitment creates pressure to show progress, and a pilot that has not met its indicators is an awkward thing to explain. The moment those criteria are waived for momentum, the sequencing has been abandoned in substance while retained in language, which is the worst of both worlds.

The unglamorous conclusion

There is a version of this story in which P2P lending is the disruptive force that finally breaks the collateral stranglehold on Nepali credit, and it happens fast.

I do not believe that version, and I do not think it is the one Nepal Rastra Bank is pursuing. The Monetary Policy commits to studying P2P based on individual credit scoring, and that qualifying phrase is not decoration. It is the design.

The likelier and better story is duller. The unique identifier arrives. The credit bureau moves from recording only failure to recording performance, and reaches past the one adult in ten it sees today. Utility and telco data make the thin-file borrower legible. Centralised KYC removes the friction of proving who you are, over and over, to institution after institution. A scoring system emerges that is reliable enough for a lender to price a borrower they have never met.

Collateral gradually loses its monopoly on trust. And in that environment, a small licensed cohort of platforms operates under caps, watched closely, with loan-level data flowing to a supervisor who has already built the capacity to read it. If they perform, they graduate. If they do not, they do not.

That is a slow path to closing a gap worth 14 percent of GDP. It is also the only path that has ever worked anywhere.

The temptation now, with the budget spoken and the mandate granted, will be to treat the infrastructure as a detail to be sorted out along the way, and the pilot as a formality en route to launch. That temptation should be named for what it is: the belief that we can have the distribution without the information, the venue without the trust, the bridge without the foundations.

We cannot. But here is the genuinely hopeful part, and it is why this moment is different from the last several.

For the first time, the foundations are actually being poured. Funded, dated, and under way.

The boring part is not the obstacle standing between Nepal and P2P lending. The boring part is the bridge. Everything else is just a website.

The views expressed here are personal and do not represent the official position of Nepal Rastra Bank.

Sources:

World Bank, Sustainable and Inclusive Finance Project: Project Appraisal Document (Report No. PADHI01141, January 2026), citing IFC and the SME Finance Forum, MSME Finance Gap (March 2025);

Nepal Rastra Bank, Monetary Policy for FY 2083/84;

Nepal Rastra Bank, Macroeconomic Report (July 2026);

Nepal Rastra Bank, Consultative Document on P2P Lending and Crowdfunding.