Thailand vs Korea — B-axis as Mirror Image

Korea's Balance-of-Power axis means 46.3 on KOSPI and 33.8 on KOSDAQ. Thailand's means 72.79 across the SET Mainboard.

The twenty-six-and-a-half-point gap is not a quality gap. Both numbers describe broken governance.

Korea's brokenness is collapsed — a single dominant mechanism applied uniformly across 2,099 listed firms,^[1] producing a B-axis that sits in the 30s and 40s with low dispersion because every chaebol-style issuer carries the same structural problem. Thailand's brokenness is decorrelated — a high mean with a standard deviation of 13.40 that decomposes across NVDR voting separation, family-cohort concentration, foreign-ownership exposure, and the form-versus-substance gap on independent directors. Four structural sources, applied to different subsets of 641 firms.

Same axis name. Mirror-image pathology. Two structurally different governance problems hiding under the same letter B.

This Note documents the comparison. It is also the bridge piece between the Korea Foundation Series (where the chaebol B-axis collapse anchors that market's Note 1) and the Thailand Foundation Series Notes 1 and 2 (where the SET Mainboard cohort distribution and the NVDR mechanism establish Thailand's distinct signature). The framework's contribution, as the comparison will show, is not that it produces the same number for the same governance problem in two markets. It is that it produces different numbers for different governance problems and captures both in the same axis architecture.

Korea: collapse through uniformity

The Korea Foundation Series anchors on a single statistic. Eighty-eight percent of the Korean listed universe (KOSPI and KOSDAQ combined) falls into the Chameleon archetype.^[2] Within those Chameleons, 89.7% are B-weak. The B-axis mean across the Korean listed universe sits near 46 on KOSPI; KOSDAQ is lower at 33.8. The mechanism is documented and well-understood: nominally-independent outside directors, nominated and refreshed under controlling-shareholder influence on a chaebol cross-shareholding lattice, function as aligned actors at the board table. Formal independence is intact on paper. Functional independence is collapsed in practice.

The compliance numbers tell the formal-side story cleanly. Korean Commercial Code §542-8 requires KOSPI-listed firms to maintain at least one-quarter outside directors, rising to one-half for firms with assets above KRW 2 trillion.^[3] Compliance is near-universal — the form is satisfied across the universe. But substantive dissent is rare. The Korea Foundation Series disclosed median outside-director dissent at approximately 11% of resolutions^[4] — meaning that in roughly 89% of board votes, all outside directors voted with the chair-and-controlling-shareholder slate. Director tenure compounds the problem: 27.3% of KOSPI firms register a zero score on the framework's outside-director-tenure indicator, corresponding to outside directors with tenure exceeding the framework's nine-year staleness threshold — long enough that the formal-rotation-and-renewal mechanism that nominal independence relies on is functionally broken. Chair-CEO role combination affects 28.9% of KOSPI firms — combining oversight and execution in the same individual.

The mechanism is uniform. Every chaebol-affiliated listed firm faces the same architecture. Every firm produces the same B-axis collapse. The framework's B-axis on KOSPI reads the collapse and outputs a low number — at 46.3, with most of the universe compressed near that mean.

Korea's B-axis is collapsed AND anti-predictive. A low number that also fails to predict events on its own, because the formal-independence-collapse mechanism is so uniform across the universe that the axis variance does not separate event firms from non-event firms. The Korea Foundation Series articulated this finding through the ablation work: the B-axis is quality, not predictor. It captures something real about the market's governance architecture, but the structural-prediction signal sits in the T-axis and R-axis where Korean cross-firm variance is larger.

Thailand: decorrelation through multi-source variance

Thailand's B-axis sits at 72.79. The number is twenty-six and a half points higher than Korea's KOSPI baseline. It is also decorrelated from event probability at the composite level — meaning the B-axis variance is real (sd 13.40, nearly twice the T-axis sd of 7.53) but does not, in aggregate, track the outcome variable as cleanly as the T-axis or R-axis do.

The four structural sources of B-axis variance in Thailand:

NVDR voting separation. Thirty-nine SET Mainboard firms have an NVDR Dilution Ratio above 15%; one hundred and ninety firms (29.6% of the universe) have at least 5% of their listed share base held in non-voting form. The framework's NVDR Dilution Ratio indicator on the B-axis produces a monotonic relationship — B-axis mean drops from 79.05 at NDR ≤ 5% to 47.06 at NDR > 70%, a 32-point spread driven by NVDR concentration alone. The mechanism is documented at length in Note 2 of this series. Korea has no equivalent instrument: KOSPI and KOSDAQ have no NVDR-style Non-Voting Depositary Receipt mechanism, and foreign capital flows through direct holdings only.

Family-cohort concentration. The framework's curated family-alias dictionary maps 33.4% of the SET Mainboard universe to 18 conglomerate clusters. The cluster-level B-axis variance is itself dispersed — CP Group's cluster mean B is 62.5 and Thai Summit's is 64.2 (drag patterns), while Mitr Phol's is 81.7 and Boonrawd's is 77.2 (lift patterns). The family cohort does not move B-axis uniformly the way the chaebol affiliation does in Korea. Some Thai family clusters drag the axis down through tenure or NVDR exposure; others lift it through governance-conscious board composition. Note 4 in this series documents the cluster-level pattern.

Independent director ratio: form versus substance. SET Corporate Governance Code 2017 requires only one-third independent directors as a minimum — a lower bar than Korea's one-quarter / one-half threshold under §542-8.^[5] Thailand's compliance against the SET minimum is high: 615 of 640 firms with identifiable board data (96.1%) clear the one-third bar. Mean independent-director ratio is 43.77%; median is 42.86%; 199 firms (31.1%) have boards at 50% or more independent. The form is satisfied. The substance — measured through dissent rate, tenure rotation, and committee-level discretion — is harder to capture from SET filings, because Thailand's board-vote disclosure regime is weaker than Korea's. The form-versus-substance gap is real but is not directly observable at the indicator level the way it is in the Korean disclosure environment.

Foreign ownership exposure. Mean foreign ownership across SET Mainboard is 37.31%, with median 42.67%; 78.2% of the universe has foreign ownership above 25%. Foreign capital presence is structurally large, and it interacts with NVDR — because foreign demand exceeds the Foreign Ownership Limit ceiling on many sectors and routes through NVDR for the marginal exposure. Korea's foreign ownership baseline on KOSPI is in a similar range historically, but the absence of an NVDR-equivalent in Korea means the foreign exposure is silent on the B-axis architecture rather than active.

The four sources interact rather than aggregate. The variance is large, some of it tracks event probability (the NVDR pathway tracks cleanly, as Note 2 documented), but the composite-level signal is decorrelated because the four sources move different subsets of firms in different directions.

Thailand's B-axis is decorrelated AND anti-predictive. A high number that also fails to predict events on its own, but for a different structural reason than Korea's collapse.

The mirror

Two pathologies. One axis label.

Korea collapses the Balance-of-Power axis through formal-independence failure on a chaebol cross-shareholding lattice. One dominant mechanism applied uniformly across 2,099 firms. B-axis mean compressed in the 30s and 40s. Most of the variance squeezed out.

Thailand decorrelates the Balance-of-Power axis through multi-source variance — NVDR voting separation, family-cohort concentration, foreign-ownership-and-FOL interaction, INED form-versus-substance. Four structural sources applied to different subsets of 641 firms. B-axis mean elevated to the 70s. Variance preserved at sd 13.40.

The numerical signature is opposite. The implication for the framework is identical: in both markets, the B-axis on its own does not predict events, and the composite-level signal sits primarily in the T-axis and R-axis variance.

The Chameleon-archetype distribution mirrors the same pattern. Korea's Chameleons are 89.7% B-weak — almost-monolithic single-source pathology. Thailand's Chameleons are 63.7% B-weak, with 31.9% R-weak, 3.6% balanced, and 0.8% T-weak — a multi-source dispersion. The dispersion of Chameleon sub-tags is itself a market-distinctive signature. Korea's Chameleon problem is monolithic. Thailand's is multi-source. The mirror runs all the way down.

The Kill Switch profile shows the same divergence at the override layer. Korea's KS triggers concentrate on a single indicator family — disclosed in the Korea Foundation Series methodology work as a near-monopoly path. Thailand's KS triggers distribute across five families: non-filing (17 firms), trading-suspension cascade (16), audit-disclaimer (10), SEC criminal complaint (8), NVDR extreme dilution and backdoor-listing-plus-loss (one each). Korea single-source. Thailand multi-source. The B-axis pattern is not an isolated finding — it is one slice of a market-distinctive structural difference that shows up across multiple framework dimensions.

What the indicator-level comparison shows

The B-axis is not a single measurement. It decomposes into structural dimensions that exist in both markets, dimensions that exist in only one, and dimensions where the same statistic carries different meaning.

Where the same statistic carries different meaning. Independent director ratio is high in both markets — Korea's near-universal §542-8 compliance and Thailand's 96.1% at the SET one-third minimum. Audit committee independence is high in both markets — Korea's §542-11 mandatory full-independent regime for large firms and Thailand's 98.3% of firms with 100%-independent audit committees.^[6] The formal-compliance numbers look similar. The structural reading does not. In Korea, high formal compliance coexists with substantive collapse measured through dissent rate and tenure. In Thailand, high formal compliance coexists with structural variance measured through NVDR and family-cluster heterogeneity. The same compliance percentage points to different governance pictures depending on which market the framework is reading.

Where the dimensions diverge. Chair-CEO combination is the unexpected differentiator. Thailand splits the chair and CEO roles in 85.7% of cases (only 14.3% combine the two), against Korea's 71.1% split (28.9% combined). Thailand splits the roles twice as often as Korea, despite being a more family-controlled market overall. This challenges the naïve assumption that family-control implies chair-CEO combination implies governance weakness. In Thailand, family-control pathology shows up via NVDR and tenure on the founder side — not via chair-CEO role combination. The role-separation discipline is structurally healthy in Thailand; the B-axis weakness sits elsewhere.

Where the dimensions are unique to one market. The NVDR row is the single sharpest cross-market differentiator. Korea has no NVDR-equivalent instrument, and the framework's NVDR Dilution Ratio indicator on the Thai B-axis has no analogue on the Korean B-axis. This is not a translation problem. It is a localization fact: the framework's per-market indicator stack is structured around the institutional features that exist in each market, and instruments that exist in only one market are measured only in that market.

The audit-firm concentration row is structurally distinctive without being unique. Korea's KOSPI audit market is approximately 80% Big-4 — a tighter concentration than Thailand's 69.9% Big-4 with a 30.1% non-Big-4 long tail. EY is the single largest auditor in Thailand at 32.0%; Korea's Big-4 split is more evenly distributed. The B-axis-adjacent signal carried by audit-firm choice is therefore different in each market: in Korea, audit firm is closer to a quality floor; in Thailand, the 30% non-Big-4 cohort is where most Kill Switch firms sit, making audit-firm choice an active discriminator. The audit concentration deep dive is a separate piece in this series.

Anti-predictivity, two ways

The framework's audit work confirms a finding that holds across both markets: the B-axis is anti-predictive of governance events at the composite level. The univariate AUC of the B-axis as a standalone classifier sits below 0.5 in both Korea and Thailand. The composite AUC does not improve when the B-axis is added under matched-control event prediction.

The reason is different in each market.

In Korea, the B-axis is anti-predictive because the mechanism is too uniform. Every chaebol-style firm has approximately the same formal-independence-collapse pattern. The B-axis variance that does exist tracks differences that do not separate event firms from non-event firms — chair-CEO combination on a 28.9% baseline, tenure on a 27.3% baseline, dissent rate on an 11% median. The signal is real, but the discrimination is poor because every firm clusters near the same B-axis value.

In Thailand, the B-axis is anti-predictive because the mechanism is too dispersed. The four structural sources of B-axis variance interact rather than aggregate. NVDR pulls the axis down on 39 firms. Family-cluster heterogeneity pushes it in different directions on 214 firms. Foreign-ownership exposure raises it through index-quality firms while masking voting asymmetry. INED-form compliance keeps the formal-side number high while the substantive variance sits below the indicator's resolution. The signal is real on each source individually — the NVDR ladder produces clean monotonicity, as Note 2 documented — but the composite combines them and the noise washes out the directional signal.

The qualitative finding is consistent across the framework's robustness checks in both markets. The specific magnitudes — the leave-one-out delta values for the B-axis on each market — are reserved for forthcoming methodology disclosure.^[7] The directional finding holds: in Korea and Thailand, the B-axis is quality, not predictor.

Why this matters for cross-market scoring

The framework operates on the principle that cross-market grade comparability is preserved through identical TBR weights (T 0.30 / B 0.30 / R 0.40) and identical archetype rules across all eight Asian markets, while per-market indicator localization captures the institutional features unique to each jurisdiction. Note 7's mirror finding is the test case for that architecture.

If a Korea-only B-axis weighting were imported into Thailand: Sino-Thai conglomerates with high formal independence and large-cap Big-4 audit infrastructure would be classified as governance-strong, despite NVDR-mediated voting separation that the Korean framework has no analogue for. Thailand's 96.1% INED compliance would dominate the B-axis signal, hiding the NVDR and family-cohort variance underneath. The result would systematically overweight family-controlled large-caps and underweight NVDR-pathway firms — the inverse of what minority-voice protection in Thailand requires.

If a Thailand-only B-axis weighting were imported into Korea: KOSPI firms with collapsed B-axis numbers would be classified as universally weak without distinguishing the formal-versus-substantive independence gap that the Korean indicator stack picks up through chair-CEO combination, tenure long-tail, and dissent-rate disclosure. The Thailand-framework's NVDR-related indicator weight would be inert across the entire Korean universe — signal lost, noise gained. The result would underdiscriminate within the chaebol universe, where every firm would look similarly bad without the framework distinguishing the architecture variants that matter for predicting events.

Cross-market governance scoring is not a translation problem. It is an indicator-localization problem. The axis names and weights stay consistent across markets to preserve grade comparability. The indicator stack varies per market to capture the institutional features that produce the axis variance in each jurisdiction. The B-axis signal in Korea is captured through the indicators that read formal-versus-substantive independence; the B-axis signal in Thailand is captured through the indicators that read NVDR exposure, family-cluster concentration, and the unique features of the SET disclosure environment. The composite reads consistently. The underlying signal is market-appropriate.

One slice of an eight-market spectrum

The Korea-Thailand contrast is one slice of a wider pattern. Each Asian market the framework covers carries a different B-axis pathology shaped by the institutional architecture of the market it operates in.

Japan's Balance-of-Power axis carries a keiretsu-shaped twin-axis residual — the lattice of cross-shareholdings that the framework reads on both the B-axis (board influence) and the R-axis (related-party perimeter). India's effective-control collapse runs through promoter pledging, where the controlling family's economic exposure is decoupled from its voting exposure through pledged-share collateral patterns. Taiwan's KY-shell pattern routes B-axis variance through offshore-incorporation transparency penalties. Hong Kong's universe distributes across five distinct ownership origins — Mainland, red-chip, family, international, cross-listed — each with its own B-axis signature. Singapore operates as the onshore hub for cross-border family cohorts whose listings sit elsewhere. The Philippines coverage is documented separately.

The framework's architectural choice was deliberate. Same axis names. Same TBR weights. Different per-market indicator stacks. The goal is not to force every market's B-axis into the same numerical signature. The goal is to capture each market's structural pathology in its own institutional terms, and to preserve composite-level grade comparability across markets that look nothing like each other underneath.

Korea collapses B-axis through uniformity. Thailand decorrelates B-axis through multi-source variance. Both arrive at the same composite-level conclusion — the B-axis on its own does not predict events — for opposite structural reasons.

The same letter B. Two different governance problems. One framework that reads both.

Korea's B-axis is collapsed (46.3 KOSPI). Thailand's is decorrelated (72.79 SET). The 26.5-point gap measures different types of broken governance.

---

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.