Form and Substance: HK INED Tenure After the 2023 Cap

Two Numbers, One Bimodal Pattern

In the cohort of 1,656 Hong Kong-listed issuers^[1] for which the framework can extract Independent Non-Executive Director tenure data with date-specificity, 52.9 percent — 876 firms — carry zero INEDs whose tenure exceeds the nine-year mark. Eighteen point six percent — 308 firms — carry a majority of INEDs over the nine-year mark. The middle bands are sparsely populated. The distribution is bimodal: a clean post-cap cohort and a binding cap-pressure cohort, with relatively few firms in between.

The Hong Kong Stock Exchange's Corporate Governance Code (Appendix C2 to the Listing Rules) introduced a nine-year tenure provision in 2023^[2], requiring enhanced disclosure and a separate shareholder resolution at the annual general meeting for the re-election of any INED whose tenure on a single board exceeds nine years. By the framework's current measurement window — the FY2024-Dec / FY2025-Dec straddle cohort, both reporting cycles post-2023-amendment — the universe-wide effect is visible. Slightly more than half of the measurable cohort has cleared its long-tenured INEDs from the independent classification entirely. Slightly less than one in five remains structurally bound by long-tenured INEDs that the provision has not yet displaced. The bimodal pattern is the post-implementation signature.

This Note examines what the bimodal distribution shows about the relationship between formal listing-rule compliance and substantive board independence in Hong Kong governance. The framework's reading is that the 2023 cap has produced a sharp form-compliance effect on the cleanest cohort while leaving the binding-pressure cohort largely intact. The architecture of that asymmetry is the substance of what follows.

The 2023 Cap and What It Compels

HKEX Listing Rule 3.10 has required, since 2012, that the board of every listed issuer include at least three Independent Non-Executive Directors and that INEDs constitute at least one-third of the board^[3]. The Listing Rule is a structural requirement: independence is defined under the Rule's specified criteria, the count and proportion are observable in continuing-disclosure filings, and substantive non-compliance triggers HKEX enforcement action.

The Corporate Governance Code (Appendix C2) operates differently. Code provisions are subject to the comply-or-explain standard, with non-compliance permitted on disclosure of the firm's reason. The 2023 amendments to Appendix C2 added a tenure-related provision: an INED who has served on the board of the same issuer for nine years or longer should be re-evaluated on independence and, if retained, should be subject to enhanced disclosure including the rationale for continued independence and shareholder approval. Issuers retaining long-tenured INEDs are required to explain the retention; issuers reclassifying long-tenured INEDs as non-independent or replacing them with new independent appointments do not.

The cap's architecture is therefore not a hard prohibition but a friction: a separate-resolution and enhanced-disclosure requirement that raises the cost of retaining long-tenured INEDs without forbidding the practice. The framework's INED tenure indicator (B-04) measures the universe-level response to this friction. The bimodal distribution observed in the framework's current measurement window is the empirical signature of how that friction has been allocated across the issuer base.

The Bimodal Distribution

Of the 1,656 firms with measurable INED tenure data, the distribution by share of INEDs over the nine-year mark falls into the following bins.

The cohort splits sharply at the endpoints. The clean cohort — 52.9 percent of measured firms — has no long-tenured INEDs at all. These are issuers whose post-2023 board composition decisions cleared the long-tenured INEDs through retirement, reclassification, or replacement. The binding-pressure cohort — 18.6 percent of measured firms, combining the 50–75% and 75–100% bins — carries long-tenured INEDs as a majority of its independent directors. Within that cohort, the 75–100% sub-bin (183 firms) is the most extreme: issuers whose entire independent slate is long-tenured.

The middle bands (over 0% and up to 50%) account for 28.5 percent of measured firms — issuers with mixed boards combining shorter-tenured and long-tenured INEDs. The middle is real but smaller than either endpoint.

Universe-level summary statistics on pct_long_tenured (the share of a firm's measurable INEDs whose tenure exceeds nine years) confirm the shape. Mean: 25.9 percent. Median: 0 percent. IQR: 0 to 50 percent. Standard deviation: 33.6 percent. The mean above the median by 26 points, with an SD comparable to the mean's magnitude, reflects the bimodal concentration. The framework's reading is that the universe is not centered on the cap; it is split around it.

Origin and Archetype Variation

The within-origin breakdown of the bimodal pattern is largely uniform across the four origins for which sufficient measurement coverage exists.

HKL and PCH — the two largest origins — converge on the clean-cohort share at 53.1 percent and 52.1 percent. The 1.0-point gap is within sampling variation. Whatever cohort-specific reduction may have followed the 2023 cap, current production measurement does not show a divergent post-cap pattern between Hong Kong-Local and P-chip cohorts. HSH (n=14) sits slightly cleaner at 57.1 percent on a small base. RCH (n=15) sits in the opposite direction with 40 percent clean and the highest mean long-tenured rate at 32 percent — Red Chip issuers, on this small base, retain long-tenured INEDs more than other origins; the small base limits inferential weight, and specific cohort drivers are not directly observable in the framework's current data. The FOR cohort is excluded from this analysis: the parser cannot extract INED tenure dates from foreign secondary issuers' HK filings with sufficient precision to compute the share.

The archetype-level breakdown shows a sharper structural pattern.

The Celestial and Hidden Gem cohorts are the cleanest — over 60 percent of each carries no long-tenured INEDs, with mean long-tenured shares around 19 percent. The Poison Apple cohort is the least clean (49.1 percent at zero long-tenured, 29.2 percent mean) — consistent with the cohort's structurally lower B-axis identified in Note 3^[4].

The Kill Switch cohort presents a counterintuitive finding: 58.5 percent clean and 22.4 percent mean long-tenured, both better than the universe baseline. KS firms are not particularly weak on INED tenure. Their KS classification is driven by other triggers — audit-opinion failure, prolonged suspension, or related-party loan exposure — not by board-tenure indicators. This is informative on what the Kill Switch override actually captures.

The HKL Poison Apple subset, where this pattern matters most quantitatively, carries the highest within-cohort long-tenured rate at 29.9 percent on 327 measured firms. The combination of high disclosure compliance with binding-pressure INED tenure is concentrated in the HKL Poison Apple population.

Form Compliance and Substantive Discipline

The form-versus-substance asymmetry visible in the INED tenure distribution is reproduced across the broader Balance of Power axis. The framework's B-axis indicator scores aggregate several sub-components measuring different aspects of board governance. Their universe-level distributions show where formal HKEX compliance is achieved and where substantive discipline lags.

The B-axis sub-component table below uses a coverage of 2,447 firms (88.4 percent of universe) — the population for which each B-axis indicator's score was successfully computed by the production parser. The tighter 1,656-firm cohort referenced earlier in this Note reflects the subset for which date-specific INED tenure measurement is feasible — the 1,656-firm subset is nested within the 2,447-firm subset (1,656 ⊆ 2,447). The broader cohort is appropriate for cross-component %-at-max comparisons; the narrower cohort is required for tenure-specific quantitative claims such as the 52.9-percent-clean and 18.6-percent-binding-pressure shares that anchor the bimodal distribution finding.

The strongest form-compliance signal is gender diversity: 62.8 percent of issuers achieve the maximum sub-component score, reflecting compliance with HKEX's 2024 single-gender-board prohibition^[5]. The INED ratio sub-component carries a moderate compliance pattern at 44.3 percent achieving the maximum, with about a third of issuers at the floor or below.

The INED tenure sub-component runs in the opposite direction. Only 7.1 percent of universe issuers achieve the maximum sub-component score on tenure discipline. The remaining 92.9 percent carry some level of tenure score below the maximum, with 39.8 percent at zero. The substantive discipline that the 2023 Code Provision's friction is designed to encourage — building an INED slate composed entirely of within-nine-year-tenure directors — has been achieved by a small fraction of the issuer universe. The clean cohort identified in the bimodal distribution above (52.9 percent of the measurable cohort) achieves zero long-tenured INEDs, but not all of these reach the maximum sub-component score; some achieve zero long-tenured but carry insufficient tenure data on remaining INEDs to score at the indicator's top band.

The framework's Listing Rule 3.10 baseline is met broadly: 71.6 percent of the 2,045 measurable cohort carry an INED ratio at or above the one-third floor, and the cross-origin range is narrow (HKL 71.4 percent to RCH 78.9 percent). HSH carries the lowest mean INED ratio in the panel at 46.4 percent — consistent with the H-share cohort's overall B-axis depression noted in Note 2 (B mean 46.1) and with the cohort's tendency to operate closer to the regulatory minimum on board independence form. The structural floor holds, but the depth above the floor is what varies.

The B-weak Mechanical Link

Note 1 identified a Chameleon archetype population of 1,445 issuers, distributed across three numerical sub-tags: [R-weak] at 48.2 percent, [B-weak] at 29.8 percent, and [T-weak] at 21.8 percent^[6]. The sub-tags identify which axis runs as the cohort's single-axis weakness. The INED tenure data in this Note maps onto that sub-tag taxonomy directly.

[B-weak] Chameleons — the cohort the framework reads as carrying its single-axis weakness on the Balance of Power axis — carry a mean long-tenured INED rate of 39.1 percent. [R-weak] Chameleons — the cohort whose weakness sits on Conflict-of-Interest Risk — carry 23.9 percent. The 15.2-point gap is structurally meaningful. A cohort that the framework's three-axis architecture identifies as B-axis-weak shows up, in the INED tenure indicator measured here, as a cohort with materially higher long-tenured INED retention.

The sub-tag taxonomy is therefore not a classification label that floats free of substantive measurement. It tracks structurally to where the binding pressure of HKEX's INED tenure cap is concentrated. The [B-weak] Chameleon population is, in part, the cohort the cap most directly bears on. That the cap has not yet displaced their long-tenured INEDs at the rate observed in cleaner cohorts is the mechanical evidence that the comply-or-explain architecture's friction has not been uniform across issuers' willingness or ability to bear it.

The [R-weak] Chameleon cohort, by contrast, sits closer to the universe baseline on long-tenured INED rate. R-weak Chameleons carry their structural weakness elsewhere — on the Conflict-of-Interest axis, through Chapter 14A connected-transaction exposure, examined in Note 5. Their B-axis profile is comparatively clean.

The Measurement Coverage Gap

The cohort analyzed in this Note is 1,656 firms — 59.8 percent of the 2,768-firm scored universe. The remaining 1,112 firms (40.2 percent) lack sufficient INED tenure-date specificity in their most recent annual report disclosure for the framework's parser to compute the share with adequate precision.

The coverage gap has two principal causes. First, directors' biographies in some current FY filings list the INED's appointment as "appointed in [year]" without month-day specificity, leaving the precise tenure unknown by a window of up to twelve months. Second, the language describing director appointment dates varies across PDF layouts, and the parser's extraction logic does not yet handle every layout variant in the universe.

The implication for this Note's findings is bounded. The bimodal distribution, the origin and archetype breakdowns, the form-versus-substance B-axis sub-component readings, and the [B-weak] Chameleon mechanical link are all observed within the 1,656-firm measurable cohort. They are not assumed to extrapolate uniformly to the 1,112 unmeasured firms. The unmeasured cohort may distribute differently; the framework's reading is that it is more likely to follow the universe pattern than to depart from it sharply, but this is not a measurement.

The pre-cap baseline (FY2022 or earlier INED tenure distribution) is also outside current production. The Phase 11 Level-2 timeseries panel was scoped to R-axis re-extraction and does not carry pre-cap B-04 measurements. The findings in this Note are therefore a post-cap snapshot, not a pre-versus-post comparison. The bimodal shape is observable in the snapshot; whether it represents a sharp departure from the pre-cap distribution, or a continuation of a pattern that existed before the cap, cannot be answered with the current data.

What the Cap Has Done, and What Remains Outside Its Frame

Three findings stand from the post-cap distribution.

First, the 2023 provision has produced a clean cohort. Slightly more than half of the measurable issuer base carries no long-tenured INEDs by the framework's current reporting cycle measurement. The form effect is real and observable. For these 876 firms, the comply-or-explain friction was sufficient to drive board-composition changes that brought the INED slate within the nine-year tenure window.

Second, the cap has not closed the binding-pressure cohort. Eighteen point six percent of measurable firms — 308 issuers — retain a majority of INEDs over the nine-year mark, and 11.1 percent retain entire INED slates that are long-tenured. For this cohort, the cap's friction has been borne or bypassed: the comply-or-explain architecture permits retention with disclosure, and these firms have made the disclosure. Whether the disclosure carries substantive justification is a separate question; the architecture compels disclosure of reasoning, not the substantive merit of the reasoning.

Third, the form-versus-substance asymmetry visible across the broader B-axis sub-components reproduces the same pattern. Form compliance — 62.8 percent of issuers at the maximum score on gender diversity, 44.3 percent at the maximum on INED ratio, 71.6 percent meeting the Listing Rule 3.10 floor — is broadly achieved. Substantive INED tenure discipline — the indicator's maximum score, achieved by issuers with no long-tenured INEDs and full tenure-data measurement coverage — is achieved by 7.1 percent of the universe. The HKEX listing-rule architecture compels form. The friction at the substantive level remains unevenly distributed, with the [B-weak] Chameleon cohort's 39.1 percent mean long-tenured rate as the framework's reading of where that friction has had the least implementation leverage.

The remaining axis the framework has not yet examined directly in this series is Conflict-of-Interest Risk. Note 5 takes up Chapter 14A connected transactions — the regulatory architecture governing the substantive transaction streams between listed vehicles and their controllers, parents, and related parties. The R-axis weakness driving the [R-weak] Chameleon spine identified in Note 1 and the high-T-low-R pattern that defines the HKL Poison Apple subset identified in Note 3 both run through the Chapter 14A regime. INED tenure is one form of binding pressure on Hong Kong governance; connected transactions are another. The architecture differs, the substantive weakness differs, but the form-versus-substance gap that this Note has examined for the board axis reappears in Note 5 for the conflict-of-interest axis. Note 6 closes the series by establishing a baseline for the climate disclosure mandate that took effect within the panel's coverage period — a regulatory architecture whose form-versus-substance dynamics will be observable in subsequent panel refreshes.

The Apex G-Score framework currently covers 2,768 Hong Kong listed companies under v2.0 calibration. INED tenure data (B-04 indicator) is parsed from issuer Corporate Governance Reports under HKEX Listing Rules and from the directors' biographies sections of annual reports; universe coverage of B-04 with date-specificity is approximately 59.8 percent (n=1,656). Pre-cap (FY2022 or earlier) INED tenure baseline is not currently in production. Listing Rule 3.10 compliance is measured under the B-03 indicator path with universe coverage of approximately 73.9 percent (n=2,045). The audit committee composition sub-component is currently affected by parser coverage limitations; its universe distribution is not used in this Note as a substantive compliance measurement.