Governance Scoring for NEPSE Companies

First published 21 Aug 2026 · Last verified 29 Aug 2026

Every serious NEPSE investor eventually develops a private list of companies they will not touch, regardless of the price-to-book multiple or the dividend yield on offer. Ask them why, and the answer is rarely a spreadsheet. It is usually a story — a rights issue that diluted minority shareholders the week before a related party got a favourable loan, an AGM that was adjourned twice without explanation, a promoter family whose shareholding quietly slipped from 51 percent to 34 percent over three years while the share price was propped up by retail enthusiasm. These stories are real, and the instincts built from them are not worthless. But instinct is not a system. It cannot be taught to a junior analyst, it cannot be back-tested, it cannot be applied consistently across the 250-plus companies listed on NEPSE, and it is vulnerable to the single greatest bias in equity investing: the tendency to excuse the governance failures of a stock you already own and to see them everywhere in a stock you do not.

This chapter closes Part IV by converting the qualitative material of the preceding chapters — promoter behaviour, board structure, related-party dealing, disclosure culture — into something that can be scored, recorded, revisited, and defended. It does not claim to import a foreign rating agency's methodology wholesale, because no such agency currently rates the governance of individual NEPSE-listed companies on an ongoing basis. What follows is an original practical framework, built specifically around what a retail investor in Nepal can actually observe from public filings, and explicitly modelled on the logic — though not the specific indicators — of internationally recognised frameworks such as the G20/OECD Principles of Corporate Governance and institutional scorecards used in comparable emerging markets. Where the chapter borrows structure from those sources, it says so. Where it is proposing something new because Nepal's disclosure environment demands a different tool, it says that too.

Lesson 23.1 — Why a Scorecard Beats a Gut Feeling

The case for structured scoring over impression is not an aesthetic preference for tidiness. It rests on four specific failures that unstructured governance judgment reliably produces, each of which has a concrete cost in a NEPSE portfolio.

The first failure is inconsistency across companies. An investor who is alert to promoter share pledging in one bank but never checks for it in another is not applying a governance view — they are applying a mood. Two commercial banks with structurally identical related-party lending patterns will get different treatment depending on which one the investor read about most recently, or which one has a chairman they personally find likeable. A scorecard forces the same six questions to be asked of every company, in the same order, every time.

The second failure is recency and narrative bias. A single well-publicized scandal — a cooperative collapse, a merchant banker's licence suspension, a bank's NRB-imposed restriction — tends to dominate an investor's governance assessment of an entire sector for months, while quieter, more persistent governance decay in a company that has stayed out of the news goes unmeasured. Scoring on a fixed schedule, from fixed inputs, corrects for this: a company's governance score should move because its disclosed behaviour changed, not because sentiment about its sector changed.

The third failure is the inability to size a decision. "I don't love this company's governance" is not information a portfolio can act on. Should the position be zero, or half the size it would otherwise be, or fully sized but subject to a tighter stop? A gut feeling gives no answer. A score that sits on a defined 0–12 scale, with defined bands, converts a vague unease into a specific, repeatable portfolio rule — discussed in Lesson 23.5.

The fourth failure is the absence of an audit trail. When a governance-driven decision goes wrong — the investor avoided a stock that then performed well, or held one that then blew up — an unstructured judgment leaves nothing behind to learn from. A scorecard, dated and dimension-by-dimension, is a record. Two years later it is possible to go back and see exactly which input was wrong: was the related-party transparency score too generous because a disclosure was taken at face value, or did the promoter shareholding trend get missed because the filing was never checked? Improvement requires a paper trail, and instinct does not leave one.

None of this argues that a scorecard replaces judgment. It structures judgment, forces it to be applied evenly, and creates the discipline of writing down why a company scored the way it did. The judgment is still the investor's. The scorecard just stops the judgment from being reinvented, inconsistently, every time a new company crosses the watchlist.

Lesson 23.2 — Design Principles for a Nepal-Buildable Scorecard

Before building the framework itself, it is worth being explicit about the constraint that shapes every dimension in it: a Nepal governance scorecard is only useful if it can be completed entirely from what a retail investor can obtain without privileged access. This sounds obvious, but it rules out most of what a global institutional scorecard — built for markets with mandatory XBRL filings, searchable board-minute archives, and independent proxy advisory research — would normally include.

The World Bank's Report on the Observance of Standards and Codes (ROSC) assessment of Nepal's corporate governance framework, benchmarked against the OECD Principles, found compliance ranging from "partially observed" to "not observed" across most categories, and identified specific structural gaps that remain directly relevant to what a scorecard can and cannot measure today: ownership disclosure is thin (listed companies report shareholding changes above certain thresholds to NEPSE but not always in an easily public form), related-party transaction rules lack the formal approval and disclosure procedures found in more developed markets, and SEBON's enforcement powers are limited — it can issue guidelines and directives but has historically relied on persuasion rather than binding secondary regulation, with enforcement action rare. This is the environment the scorecard must be built for, not the environment an investor might wish existed.

Since that assessment, Nepal's framework has moved incrementally. The Companies Act 2063 (2006) requires public companies to maintain a minimum board size, include at least one independent director with no material relationship to the company beyond the directorship, include at least one female director, and have at least one director ordinarily resident in Nepal. The Bank and Financial Institutions Act (BAFIA) 2073 adds board committee requirements and "fit and proper person" criteria specifically for banks and financial institutions. SEBON, for its part, has periodically issued directives requiring listed companies to submit a standardised annual corporate governance disclosure alongside audited financials — covering board composition and conduct, risk management and internal controls, and organizational structure and staffing, submitted in a uniform format designed for comparability across companies.

These are genuine, checkable disclosure obligations, and they are precisely the raw material a Nepal scorecard should be built from — annual reports, AGM notices and minutes, the SEBON-mandated corporate governance report, NEPSE's disclosure portal, and the audited financial statements themselves. Four design principles follow from this reality.

Principle one: score only what is disclosed, not what is assumed. If an annual report does not state the number of board meetings held, the scorecard records that as an absence of disclosure — itself a governance data point — rather than guessing at a number.

Principle two: prefer trend over snapshot wherever the data allows it. A promoter holding 45 percent of shares tells you very little in isolation; a promoter holding that fell from 58 percent to 45 percent over three annual reports tells you a great deal, and the direction is more informative than the level.

Principle three: separate what a company is legally required to disclose from what it discloses voluntarily. A company that goes beyond the SEBON-mandated minimum — publishing related-party transaction schedules in more detail than required, or disclosing director attendance at board meetings — is signalling something about its governance culture that a company doing the legal minimum is not, even if both are technically compliant.

Principle four: build the scorecard to be completed from the same four or five source documents every time, so that scoring one company does not require a different research process than scoring the next. This is what makes a scorecard usable at scale rather than as a one-off boutique exercise for a single favourite stock.

Lesson 23.3 — The Six-Dimension NEPSE Governance Scorecard

The framework below organises governance into six dimensions, echoing the broad structure used by international scorecards — shareholder treatment, disclosure quality, and board responsibility are recognizable categories from the OECD Principles and from institutional scorecards such as those used by Indian proxy advisory firms — but the specific indicators are chosen because they are things a NEPSE investor can actually check, unlike indicators requiring board minutes access, private proxy advisor research, or regulatory non-public filings.

Each dimension is scored 0, 1, or 2. Total possible score across six dimensions is 12.

Two design notes on this table matter for how it should actually be used. First, the related-party transaction dimension and the promoter shareholding trend dimension are not independent of each other in practice — a promoter quietly reducing their stake while related-party lending to promoter-linked entities increases is a specific, recognizable pattern, and an investor using this scorecard should read the two rows together, not just sum their scores mechanically. Second, the audit quality dimension deliberately treats an unusually long, unrotated auditor tenure as a caution flag even in the absence of a formal rotation mandate, because Nepal's Companies Act framework does not impose the kind of hard auditor-rotation ceiling found in some other jurisdictions — which means an auditor relationship can run for many years with no external circuit-breaker, and the burden of noticing that falls on the investor rather than the regulation.

Once each dimension is scored, the six scores are summed into an overall grade band:

Lesson 23.4 — Worked Example: Scoring a Hypothetical Commercial Bank

Consider an illustrative, unnamed commercial bank — call it "Bank X" — of the kind commonly found in the NEPSE "A" category. The example is constructed to show how the scorecard is actually applied to a real annual report, not to describe any specific listed institution.

Board independence and composition. Bank X's annual report lists a nine-member board: five representing the promoter group, two representing public shareholders, one professional director appointed per NRB norms, and one designated as independent with a one-paragraph note on the criteria used (no conflicting business interest, no relative of a promoter). This clears the bare statutory minimum and discloses selection criteria for the independent seat, but the board remains promoter-dominated in composition. Score: 1.

Related-party transaction transparency. The notes to the financial statements disclose loans and facilities extended to entities where a director or promoter holds a substantial interest, itemized by counterparty name, outstanding balance, and interest rate, with a note that the audit committee reviewed the schedule. This is a meaningfully more detailed disclosure than the bare aggregate figure many smaller companies provide. Score: 2.

Promoter shareholding trend. Comparing NEPSE shareholding disclosures across the past four annual reports, the promoter group's stake has moved from 51 percent to 49 percent, entirely explained by a mandatory public share issuance requirement rather than any promoter sale. The decline is real but fully attributable to a disclosed, ordinary corporate action. Score: 2.

Disclosure timeliness and quality. The AGM was held within the statutory window in three of the last four years, with one year's AGM delayed by roughly two months with a publicly stated reason (delayed audit sign-off pending an NRB inspection query). Quarterly filings were consistently on time. Score: 1, reflecting the one lapse.

Dividend and rights issue history. The bank has paid a mix of cash and stock dividends over the review period, with a rights issue three years prior priced at par and timed to coincide with a capital adequacy requirement, disclosed clearly in the AGM notice with the rationale explained. No pattern of dilutive timing against minority shareholders is evident. Score: 2.

Audit quality and auditor tenure. The bank has used the same audit firm for the past six years, with clean opinions throughout, but the annual report does not describe any audit committee deliberation on whether to rotate the auditor. The tenure is long enough to warrant a note but the opinions themselves show no qualification. Score: 1.

Summing these: 1 + 2 + 2 + 1 + 2 + 1 = 9, placing Bank X at the top of the "B" band — adequate governance, with two specific, named items to monitor going forward (promoter-dominated board composition, and the absence of visible auditor-rotation deliberation), rather than either an unqualified pass or a disqualifying red flag. This is precisely the kind of nuanced, defensible output the scorecard is meant to produce: not a verdict of "good" or "bad," but a specific, dated record of where this bank's disclosed governance stands and which two items should be re-checked at the next annual report.

Lesson 23.5 — Using the Score in Practice

A governance score only earns its place in the process if it changes a decision. There are three distinct, non-exclusive ways to put it to work.

As a position-sizing input. A simple, defensible rule ties maximum position size to grade band: an "A" or "B" company is eligible for a full position sized on financial and valuation merits alone; a "C" company has its maximum position size capped — for instance, at half of what the financial analysis alone would justify — regardless of how attractive the valuation looks; a "D" company is excluded from new purchases entirely, with existing holdings reviewed for exit rather than added to. This converts governance from a vague qualifier into a hard constraint on portfolio construction, which is the only place governance judgment reliably survives contact with a tempting valuation.

As a red-flag screen at the point of first research. Before any DCF is built or any ratio is compared to sector peers, running the six-dimension score on a new name takes under an hour from public filings and can eliminate candidates before time is spent on deeper financial modelling. A stock that scores a 0 on related-party transparency and a 0 on promoter shareholding trend simultaneously — related lending rising while the promoter quietly exits — is a combination worth an automatic pass regardless of how cheap the multiple looks, because it is a classic precursor pattern to value destruction for minority shareholders.

As a factor in valuation discount or premium. For companies that clear the position-sizing bar but land in the "B" or "C" band rather than "A," a deliberate valuation discount is a more honest way to express governance concern than an outright exclusion. A bank trading at 1.1x book that would otherwise justify 1.3x book on pure return-on-equity and growth grounds, but whose governance score sits at a C, might reasonably be capped at a target of 1.0–1.1x book precisely because of the governance discount — the market's own skepticism about weak-governance names, reflected in a persistently lower multiple, is not a mispricing to arbitrage but a rational discount that a governance scorecard makes explicit and repeatable rather than something felt only vaguely.

The score should also be re-run on a fixed cadence — at minimum, once per year after the annual report is published, and again after any material rights issue, merger, or promoter transaction — rather than only when a scandal draws attention to a name. The entire value of the framework lies in it being applied evenly and on schedule, not selectively when suspicion is already aroused.

Lesson 23.6 — What This Scorecard Cannot Catch

Any DIY governance scoring system built for Nepal must be used with a clear-eyed view of its limits, and those limits trace directly back to the same enforcement and disclosure weaknesses noted in the World Bank's assessment of the framework: a scorecard built from disclosed information is only as reliable as the disclosure regime that produces it, and SEBON's limited enforcement capacity means non-compliant or superficial disclosure carries little practical consequence for the company producing it.

The scorecard cannot detect a related-party transaction that is simply never disclosed. Nepal's rules do not impose the kind of proactive, standardised related-party approval and disclosure regime found in more mature markets, so a promoter-linked entity that supplies a listed company, or borrows from a bank the promoter also controls, may not appear anywhere in the annual report if the company chooses a minimal reading of its obligations. A "2" score on this dimension means nothing troubling was found in the disclosure — it cannot mean nothing troubling exists.

The scorecard cannot see through nominee shareholding. A reported decline or stability in "promoter" shareholding is only as accurate as NEPSE's and the company's own classification of who counts as a promoter; shares held through family members, associated companies, or nominee arrangements that are not formally classified as promoter holdings can mask the real trend in controlling-family ownership, in either direction.

The scorecard cannot verify the substance behind a disclosure, only its presence. An independent director "meeting disclosed criteria" on paper may still be a long-standing family friend of the chairman in practice — something no public filing will state and no scorecard row can capture. Similarly, an audit committee "reviewing" a related-party schedule, as stated in a footnote, is not verifiable evidence that the review was substantive rather than a formality noted to satisfy the SEBON reporting template.

The scorecard is backward-looking by construction. It scores what has already been disclosed in a completed annual report, which means it will always lag a governance deterioration that is happening in real time — a related-party loan extended in the current fiscal year will not appear in the scorecard until the next annual report is published, potentially a year or more later.

The scorecard cannot substitute for financial forensic analysis. A company can score well on every governance dimension in this chapter while still carrying financial red flags — aggressive revenue recognition, understated provisioning, or asset quality issues — that belong to the financial-statement analysis covered elsewhere in this book, not to governance scoring. The two lenses are complementary, not interchangeable; a high governance score is not a certificate of financial health.

Finally, the scorecard is only as good as the discipline applied in filling it out. Its greatest practical risk is not a flaw in the framework itself but the temptation to score generously a company the investor already wants to own, and harshly one they have already decided to avoid — the exact bias in Lesson 23.1 that the scorecard was built to correct. Guarding against that requires scoring before forming a view on valuation, not after, and revisiting old scores honestly when new annual reports arrive rather than only when a new scandal makes the exercise unavoidable.