NVDR — Where Voting and Ownership Come Apart

On the SET Mainboard, you can own a company without being able to vote it.

Thirty-nine listed firms have an NVDR Dilution Ratio above 15% — a structural minority-voice-reduction layer that exists in none of the other seven markets the Apex G-Score framework covers. One hundred and ninety firms — 29.6% of the scored universe — have at least 5% of their free float held in non-voting form. The mechanism is twenty-five years old, embedded in SET's market plumbing, and almost invisible to anyone reading the standard shareholder register.

This Note documents how the mechanism works, what the framework measures, and why Thailand's Balance-of-Power axis ends up at 72.79 — twenty-six points above Korea's chaebol-collapse number — while still being decorrelated from event probability at the composite level. Same axis name. Mirror-image pathology.

What an NVDR is

A Non-Voting Depositary Receipt is a Thai-only equity-related instrument issued by Thai NVDR Co. Ltd., a wholly-owned subsidiary of the Stock Exchange of Thailand. The instrument was introduced in March 2000 as a structural workaround to the Foreign Ownership Limit ceilings imposed sector-by-sector under the Foreign Business Act and the Financial Institutions Business Act.^[1]

The mechanics are clean. A foreign investor who buys an NVDR receives the full economic exposure of the underlying SET-listed share — dividends, capital appreciation, rights issuances all pass through. But the legal voting right attaches to Thai NVDR Co. itself, which by long-standing operating practice abstains from voting at issuer general meetings, with limited exceptions. NVDRs trade on SET alongside the underlying common shares. Their issuance volume per issuer is uncapped — the constraint is foreign-investor demand, not issuer permission.

The historical bargain was deliberate. Thai issuers wanted to attract cross-border equity capital. Foreign investors wanted exposure to Thai growth. The Foreign Business Act capped direct foreign holdings at 49% (general), 30% (utilities), or 25% (banking).^[2] The NVDR's economic-without-political design solved both sides — at the cost of a structural minority-voice asymmetry that has accumulated quietly over twenty-five years.

How it shows up on the Balance-of-Power axis

When a meaningful share of an issuer's free float is held via NVDR, the universe of voting shares contracts relative to the universe of economic-interest shares. The controlling-shareholder cohort's voting concentration is therefore higher than the published shareholding percentage suggests, because the NVDR layer is silent at the ballot.

The arithmetic is straightforward. A 51% controlling stake on a register where 20% of the listed share base is held via NVDR — leaving 80% of the share base actively voting — translates to roughly 64% effective voting concentration once NVDR abstention is netted out. The published number understates the true voting amplification by thirteen percentage points.

The framework captures this as the NVDR Dilution Ratio (NDR): NVDR holdings divided by free float. The indicator is publicly disclosable; the exact weight assigned to NDR within the Balance-of-Power axis is proprietary. What is not proprietary, and what answers the question of whether the indicator carries signal, is the relationship between NDR and the B-axis score itself:

NDR bucket	n	B-axis mean
≤ 5%	257	79.05
5–15%	152	69.70
15–30%	32	61.00
30–70%	5	51.20
> 70%	1	47.06

Monotonic. Every step up the NDR ladder produces a step down on B-axis mean. From 79.05 at the lowest exposure bucket to 47.06 at the highest — a 32-point B-axis spread, structurally driven by NVDR concentration alone.

This is publication-grade evidence that the framework's Balance-of-Power axis is materially shaped by NVDR exposure and not by formal-independence variables alone. NVDR is not a footnote in the Thai governance score. It is a primary B-axis driver.

Distribution: tail and scope

The NDR cohort can be read two ways — as a tail (where the extreme values cluster) or as a scope (how many firms have any meaningful NVDR exposure at all).

The tail reading:

Threshold	n firms	% of NDR-computable	% of universe
NDR ≥ 5%	190	42.5%	29.6%
NDR ≥ 10%	90	20.1%	14.0%
NDR ≥ 15%	39	8.7%	6.1%
NDR ≥ 20%	21	4.7%	3.3%
NDR ≥ 25%	10	2.2%	1.6%
NDR ≥ 30%	6	1.3%	0.9%
NDR ≥ 50%	1	0.2%	0.2%
NDR ≥ 70%	1	0.2%	0.2%

Of the 641 SET Mainboard firms in the scored universe, NDR is computable for 447 — meaning the firm has an identifiable NVDR position somewhere in its top-30 shareholder register.^[3] The remaining 194 firms have no NVDR row visible at all, which is itself a signal: NVDR exposure is a foreign-investor-interest proxy, and the absence of NVDR holdings often marks small-cap, low-foreign-attention issuers that cross-border capital has not found yet.

The scope reading is the broader one: 190 firms with NDR ≥ 5% — three out of every ten SET Mainboard firms have at least 5% of their free float in non-voting form. NVDR is a common exposure on this market, not a rare one.

The tail of the distribution is the editorial cohort. The five firms with the highest NDR ratios:

Symbol	NDR	Archetype	Grade
F&D (Food and Drinks PCL)	113.4%	KS	KS
THIP	44.8%	Poison Apple	C
ROCTEC	40.3%	Poison Apple	C
TRUE	35.0%	Chameleon	B
GBX	31.7%	Chameleon	C

F&D's ratio above 100% is the structural-vacuum extreme. NVDR holdings are uncapped per issuer; foreign-investor demand can accumulate NVDR positions beyond the published free-float number, because free-float is a discrete reported figure while NVDR aggregation has no ceiling. When the ratio crosses 100%, the published free-float has effectively collapsed as a meaningful denominator — the entire non-controlling layer of the firm sits in non-voting form.

Archetype skew at the tail

Within the 39-firm cohort at NDR ≥ 15%, archetype distribution diverges sharply from the universe baseline:

Archetype	NDR ≥ 15% (n=39)	Universe baseline	Multiple
Celestial	5.1%	17.5%	0.3×
Hidden Gem	12.8%	34.5%	0.4×
Chameleon	61.5%	38.7%	1.6×
Poison Apple	17.9%	3.1%	5.7×
KS	2.6%	6.2%	0.4×

The headline number is the Poison Apple multiple. Among firms with NDR ≥ 15%, Poison Apple share is 5.7 times the universe baseline. Celestial share, conversely, is one-third of baseline. High NVDR concentration is a strong structural marker for the transparent-facade-with-structural-weakness archetype.

The mechanism is not subtle. A firm can score well on Transparency — bilingual disclosure, on-time filing, audit-opinion clean — and simultaneously sit on a shareholder register where most of its non-controlling float has voluntarily surrendered its vote. The T-axis sees disclosure quality. The B-axis sees the voting-power asymmetry. The composite reads as Poison Apple: the mask is in front, the structural problem is behind.

The KS-04 trigger

The framework includes a Kill Switch trigger for extreme NVDR-mediated dilution, mapped in the public taxonomy as KS·04 Concentration & Extraction. The exact NDR threshold that activates the trigger is proprietary. The qualitative framing is straightforward: when NVDR-mediated voting separation reaches a structural extreme, the firm is routed out of the archetype taxonomy entirely and into the Kill Switch override layer.

One firm currently sits in this category: Food and Drinks Public Company Limited (F&D, FD.BK), with an NDR of 113.4%.

A clarification matters here, because two firms with similar mechanisms exist in the framework's record and the distinction is easy to miss. F&D is currently active and Kill Switch flagged under the NVDR-dilution pathway.^[4] MORE Return Public Company Limited (MORE.BK) is a separate firm, delisted following its November 2022 episode, and is not in the current scored universe. MORE is a retrospective LAB-validation case — the framework's flagship demonstration that the NVDR-pathway signal can be detected before an event materializes. F&D is the real-time current trigger. Both firms exhibit the same structural signature; one closed, one open.

Family conglomerate clusters and NVDR

The framework's curated family-alias dictionary maps 18 conglomerate clusters across 214 firms. Cross-tabulating those clusters against NVDR exposure reveals a pattern worth recording.

Selected clusters by mean NDR (13 of 18 displayed):

Cluster	n with NDR data	Mean NDR	n at NDR ≥ 15%
CP Group	6	17.16%	4 / 6
PTT state-cohort	51	12.15%	15 / 51
BDMS	2	11.58%	0 / 2
Thai Summit	6	9.17%	1 / 6
Thai Union	5	7.23%	1 / 5
Bangkok Bank	21	6.48%	2 / 21
TCC / ThaiBev cohort	3	5.93%	0 / 3
Boonrawd	14	5.59%	0 / 14
Mitr Phol	4	4.54%	0 / 4
Central	10	4.16%	0 / 10
KBank / Lamsam	8	3.04%	0 / 8
King Power	6	2.47%	0 / 6
Saha Pathanapibul	3	1.52%	0 / 3

CP Group sits at the top. The cluster's mean NDR of 17.16% is more than double the universe average, and four of the six CP-cluster firms with NDR data have crossed the 15% threshold. CP is, by NVDR proxy, the most foreign-investor-exposed Sino-Thai conglomerate on the Mainboard — the cohort-wide pattern, not a single-firm artifact.

The PTT state-cohort sits in the middle range at 12.15% mean NDR with 15 firms above the 15% threshold. Large-cap state-enterprise hybrids attract foreign-investor demand proportional to their index weight, and NVDR is the conduit that demand flows through.

A specific aggregate-only note on the royal-adjacent institutional shareholding cohort: the five clustered firms in this cohort that have NDR data show a mean NDR of 6.44%, and none of the five sit above the 15% threshold. Specific firm identification is outside the scope of this Note; the cohort-level reading is the observation.

The lower end of the cluster table is equally informative. Saha Pathanapibul, King Power, KBank/Lamsam, Central, Mitr Phol — predominantly domestic and family-held, with minimal foreign NVDR uptake. These are clusters that have not opened structurally to cross-border passive flows in the way CP and PTT have. The B-axis dispersion within these clusters comes from sources other than NVDR — long director tenure, chair-CEO combinations, related-party perimeter issues — which other Notes in this series will address.

Foreign-parent firms: the counter-pattern

The 20 foreign-parent-controlled firms in the universe (DELTA, BBGI, BCPG, BGRIM, and 16 others) show a counter-intuitive NVDR profile: mean NDR of 7.04% — below the universe average.^[5] The mechanism is structural. Foreign-parent firms typically already hold their controlling stake at or near the Foreign Ownership Limit ceiling through direct (non-NVDR) holdings. Marginal NVDR uptake by additional foreign investors is therefore smaller, because the FOL headroom they would normally use has already been consumed by the parent.

NVDR exposure clusters, in other words, in firms where foreign demand is high but direct-foreign-holding capacity is constrained — the family-controlled and dispersed cohorts where foreign appetite outstrips direct-holding allowance. The foreign-parent cohort is where foreign capital has entered the front door. The high-NDR cohort is where foreign capital has come in through the side.

Korea's collapse versus Thailand's decorrelation

Korea's chaebol architecture produces a Balance-of-Power axis mean near 46. The mechanism is documented: nominally-independent directors, nominated and refreshed under controlling-shareholder influence, function as aligned actors at the board table. Formal independence is intact on paper. Functional independence is collapsed in practice. The framework's B-axis reads the collapse and outputs a low number.

Thailand's B-axis mean is 72.79. The number is twenty-six points higher. It is also decorrelated from event probability at the composite level — meaning the B-axis variance is real but does not, in aggregate, track the outcome variable as cleanly as the T-axis or R-axis do. The qualitative finding is robust across the framework's robustness checks; the specific magnitude of the decorrelation is reserved for forthcoming methodology disclosure.

Two pathologies. One axis label.

Korea's B-axis is collapsed. Thailand's B-axis is decorrelated. Korea's signal is concentrated and uniform; Thailand's signal is diffused across NVDR voting separation, family-cohort concentration, multi-decade director tenure on Sino-Thai boards, and chair-CEO co-incidence patterns. The framework's response is structural: in Korea, the B-axis is a heavy lifter that drives Chameleon dominance. In Thailand, the B-axis must be read in conjunction with T and R, not as a stand-alone verdict.

Note 7 in this series documents the comparison in full. For Note 2's purposes, it is enough to know that the B-axis the framework reports for SET Mainboard is not the same B-axis it reports for KOSPI.

What MORE Return showed in advance

In November 2022, MORE Return's stock collapsed in a single-session market-manipulation episode that forced SET's clearing-system stress controls to engage.^[6] The episode triggered SEC enforcement against the individuals involved, and the firm was subsequently delisted. By any standard accounting, this was an event the framework would want to detect in advance.

It did. At the framework's June 2022 pre-event reconstruction date — five months before the November episode — MORE's NVDR Dilution Ratio sat at approximately 65%. Its pre-event composite scored 18.3 points below the matched-control mean, with the Balance-of-Power axis as the dominant gap-driver.^[7] The mechanism the November episode exploited — coordinated price action through a thinned voting-rights universe — is the same structural pathway the NDR indicator reads.

NVDR-mediated voting separation, at extreme levels, does not just compress minority voice. It creates a structural vacuum that becomes itself a manipulation vector — because a price-impact action against a stock whose voting universe has already contracted to the controlling cohort encounters proportionally less institutional-investor pushback than it would on a register with active dispersed voting. The vacuum is the vulnerability.

Case 2 in this series treats the full MORE sequence at deep-dive length: the pre-event signal, the event mechanics, the post-event regulatory response. For Note 2, MORE is the proof-of-concept that NVDR is not just a Balance-of-Power input. In its extreme form, it is an event predictor.

What the framework reads, what it does not

The NDR indicator measures the share of free float held in non-voting form. It does not measure the shareholders behind the NVDR line — those identities are aggregated at Thai NVDR Co. and not disclosed at the issuer-register level. It does not measure abstention behavior within the limited exceptions where Thai NVDR Co. does vote. It does not capture the secondary effects of NVDR-mediated voting separation on minority-shareholder coalition formation, which would require event-study-level analysis the framework's structural-prediction layer does not currently perform.

What the framework reads is the structural fact: that on the SET Mainboard, ownership and voting come apart, and the gap between them is measurable, monotonic in the B-axis, and concentrated in 39 firms whose archetype distribution is dominated by Poison Apple at 5.7× the universe rate.

NVDR is the Thailand-distinctive mechanism. It is the reason the B-axis mean is high. It is the reason the B-axis is decorrelated from event probability at the composite level. And it is the reason the B-weak Chameleons identified in Note 1 number 158 — a structural cohort, not a measurement artifact.

The mechanism is twenty-five years old. The cohort is current.

Monotonic B-axis decline across NVDR exposure buckets. From 79.05 at ≤5% to 47.06 at >70% — a 32-point structural spread.

---

The Apex G-Score framework currently covers 641 SET Mainboard listed companies as of the April 2026 production snapshot. Underlying data: FY2025 cross-section. Scoring under TBR v2.0 weights: Transparency 0.30, Balance of Power 0.30, Conflict-of-Interest Risk 0.40.