Skip to content

FORENSIC ACCOUNTANT REPORT — THG (Thonburi Healthcare Group PCL)

Scope warning. The data package contains no income statement, balance sheet, or cash-flow line items for THG. The SET shareholders and filings pages 404'd; the SET profile/financial pages are JS-shelled and unparseable. The only quantitative financials in the package are four headline figures lifted from the dossier (FY25 revenue, FY25 NI, FY24 NI, Q3-25 numbers) plus the RAM placement mechanics. This report is therefore a structural red-flag map, not a quantified forensic teardown. Anywhere I cannot calculate, I say so. I have not fabricated ratios I could not derive.

1. Overall accounting quality grade: D (provisional)

The "D" is driven primarily by governance / fraud overhang, not by any specific accruals manipulation I can prove from the package. With statements in hand the grade could move either way.

  • Earnings quality: 2/10 — FY25 net income +฿96m on ฿9.10bn revenue = ~1.06% net margin. The entire swing from −฿1.76bn to +฿96m is dossier-attributed to interest savings, not operating improvement. Revenue is declining (−4.0% FY25; −8.3% Q3-25 YoY). Earnings are a rounding error around break-even — definitionally low quality.
  • Cash conversion: n/a (no cash-flow statement in package) — provisional 3/10 on the prior that a hospital group running a ฿1.76bn loss in FY24 that needed a ฿6.28bn recap was almost certainly cash-negative or barely positive at OCF level pre-recap.
  • Balance sheet health: 4/10 — improved materially via recap (debt down, interest cut >฿300m/yr per dossier), but post-recap net debt is unverifiable in this package. Share count up ~111% — equity dilution, not deleveraging via earnings.
  • Disclosure transparency: 1/10 — SET shareholders page 404. SET filings page 404. No 56-1, no MD&A, no Form 59-2 in the package. The founder is a fugitive facing a ฿16.1bn fraud case. Disclosure scores cannot get much worse than this combination.

2. The 10-year cash-vs-earnings picture

Cannot compute. No operating cash flow, no net income time series beyond FY24 (−฿1.76bn) and FY25 (+฿96m). What the package does tell me:

  • FY24 reported NI was −฿1.76bn on a ฿9.10bn-ish revenue base — that is a ~−19% net margin, an extraordinary loss for a hospital operator. Without the income-statement detail I cannot tell how much was impairment / one-off / Boon-related provisioning vs operating bleed. This is the single most important missing piece. A −฿1.76bn loss followed immediately by a +฿96m profit is exactly the pattern where "big bath" accounting needs to be ruled out: did FY24 absorb every conceivable write-down (receivables, goodwill, intangibles, related-party loans to fugitive-founder vehicles) so FY25 could print positive?
  • FY25 NI of +฿96m vs an interest saving "of >฿300m/yr" implies that absent the recap, FY25 would still have been a loss. Operating earnings have not turned.

3. Working capital diagnosis

Cannot compute receivables days, inventory days, or payables days — no balance-sheet detail in package.

Specific concerns that need verification from 56-1: - Hospital receivables from Thai government schemes and private insurers typically run 60–120 days; collection extensions are a common margin-management lever. - The founder allegedly ran fraudulent fake hospital projects in Laos/Vietnam (Thai PBS World, 30 Nov 2024). Any inter-company receivables, advances, or "investments in associates" flowing to those vehicles are the #1 thing a forensic reviewer would chase. Quantum: unverifiable here.

4. Revenue quality

  • Concentration: revenue split by hospital (Thonburi 1, Thonburi 2, Trang, Jin Wellbeing, Healthiva) not in package. Dossier mentions Trang Hospital "supporting additional income of ฿500m/yr" (Money & Banking, 9 Jul 2024) — ~5.5% of FY25 revenue, modest contribution.
  • Related-party share: unverifiable. This is the biggest gap. With a fugitive founder and his family historically controlling the group, related-party transactions in the notes to the financials are mandatory reading. Not in package.
  • Recognition timing: cannot test. No quarterly granularity beyond Q3-25.
  • Trajectory: revenue −4.0% FY25, Q3-25 −8.3%. The decline is accelerating, which means the +฿96m profit was achieved while the top line was shrinking — i.e. all earnings improvement came from interest line, none from operating leverage.

5. Margin trajectory & one-offs

  • FY25 net margin: ~1.06%. Peer benchmark per dossier: BDMS/BCH "double digits." Gap to peers: ~10pp at NI level.
  • Q3-25 net margin: 0.2% (฿5.12m / ฿2.32bn). One bad quarter from another loss.
  • One-offs: FY24's −฿1.76bn loss is the natural location for any non-cash impairments (goodwill, Healthiva launch costs, write-downs of Vietnam/Laos JV exposure, Boon-related legal provisions). Without the P&L breakdown I cannot quantify how much of the FY24-to-FY25 swing is real and how much is base-effect from a kitchen-sinked prior year.

6. Capex / depreciation / asset base

  • Dossier flags "THG set to spend ฿5bn on projects 2023–25" (Bangkok Post, 17 Mar 2023) and "฿1–1.5bn on three cancer centres" (Bangkok Post, 22 May 2024). That is a heavy capex programme for a group with FY25 revenue of ฿9.1bn (~55–70% of one year's revenue planned over three years).
  • Question: was this capex actually executed, partially executed, or paused under financial distress? Unverifiable here. The Jin Wellbeing elderly-care project (VOA, 11 Apr 2018) and Healthiva launch (3 Jul 2024) are both capital-hungry; whether they are generating returns or sitting as under-utilised PP&E (depreciation drag with no revenue) is the central operating question — and the package gives no answer.
  • D&A vs capex ratio: uncomputable.

7. Off-balance-sheet & leverage red flags

  • Pre-recap debt: dossier cites "combined loans/debentures ~฿9.6bn." RAM raise gross ฿6.28bn used to repay debt. Exact post-recap net debt: unverifiable (the data audit explicitly flags the "฿5.9bn repaid" figure as unconfirmed).
  • Insider share pledges: status unknown — and given a fugitive founder with ~4–8.8% residual stake plus assets under Thai AMLO seizure proceedings (SCMP, 5 Jan 2025: "Thailand moves to seize tycoon's assets in US$350m scam"), pledged-share risk has to be assumed material until disproved. This is a CRITICAL gap. Whether the Vanasin family's residual shares are pledged, frozen, or subject to forced sale by creditors / authorities materially affects free float and the risk of an overhang.
  • Financial guarantees to founder-related vehicles (Laos/Vietnam "fake hospital projects" per Thai PBS World, 30 Nov 2024): must be assumed possible until notes are reviewed. This is the single most important off-balance-sheet exposure to investigate.
  • Operating leases: not disclosed in package.

8. Auditor & policy changes

  • Auditor identity: not in package. SET filings page 404'd.
  • Restatements: not in package. Given the scale of Boon-era alleged fraud (฿16.1bn estimated damages, charges filed by AG) and the magnitude of the FY24 loss, the probability of a restatement of historical THG financials is non-trivial. Cannot confirm.
  • Board changes confirmed in dossier: Sam Tanskul (independent director, eff. 20 Feb 2025), Paradorn Leosakul (eff. 31 Mar 2025), Dr Linda Kraivit resigned 31 Jan 2025. This is consistent with a post-acquisition governance overhaul — generally a positive forensic signal, since new boards typically demand clean books before signing the next set of accounts.

9. Quantitative red-flag scores

  • Beneish M-Score: NOT COMPUTABLE. Requires 8 ratios (DSRI, GMI, AQI, SGI, DEPI, SGAI, LVGI, TATA). None of the inputs (receivables, COGS, gross margin, total assets, depreciation, SG&A, total liabilities, accruals) are in the package. Any number presented here would be fabricated.
  • Altman Z-Score: NOT COMPUTABLE. Requires working capital, retained earnings, EBIT, market cap (computable: ~฿7.30 × shares outstanding — but exact share count post-PP+RO is "~1.79bn" per dossier arithmetic, not a filed number), total assets, total liabilities, sales. Missing 4 of the 5 inputs.
  • Sloan Accrual Ratio: NOT COMPUTABLE. Requires net income (have FY25), operating cash flow (don't have), and average total assets (don't have). Even the directional sign cannot be established.

The auditor's DO-NOT-FABRICATE list explicitly forbids inventing these. I am honouring that.

10. Top 5 specific concerns, ranked

  1. Founder-fugitive related-party / receivables exposure (the Boon overhang on the books). What it is: Dr Boon allegedly ran fraudulent schemes using fake hospital projects in Laos/Vietnam (Thai PBS World, 30 Nov 2024); AG charges + ฿16.1bn estimated damages; founder fled Sept 2024. Why it matters: there is a non-trivial probability that THG's pre-2024 balance sheet contained receivables from, guarantees of, or "investments in" founder-related vehicles that may need to be written off and may, under Thai AMLO proceedings, become legally contested. The FY24 −฿1.76bn loss may already contain some of this — or may contain only part of it. Fair-value discount: 20–30% until notes confirm all founder-related exposures are written down and no residual guarantees exist.

  2. Disclosure opacity (the SET pages and filings I cannot see). What it is: shareholders 404, filings 404, no MD&A, no 56-1 in package. Why it matters: I cannot independently verify the RAM 49.99% stake, the Vanasin residual ~4.2%, post-recap net debt, related-party note, contingent liabilities, share-pledge status, or auditor identity. Fair-value discount: 10–15% information-asymmetry haircut until primary docs are read.

  3. FY24 "big bath" risk vs FY25 thin profit. What it is: −฿1.76bn → +฿96m swing on declining revenue. Why it matters: classic pattern where a new controlling shareholder (RAM) takes every possible write-down in the year before/during acquisition so subsequent reported earnings look better. Doesn't make FY25 numbers wrong — does make them a poor base for extrapolation. Fair-value discount: 10% on quality-of-earnings; don't capitalise the FY25 EPS at peer multiples.

  4. Capex commitments vs operating cash generation. What it is: ฿5bn 3-year capex plan (2023–25) + cancer-centre and Healthiva builds, against ~1% net margins and declining revenue. Why it matters: if the capex was executed, depreciation is now eating margin without revenue payoff. If it was paused, there's a stranded-PP&E impairment risk waiting. Fair-value discount: 5–10%.

  5. Share-pledge / forced-sale risk on the residual Vanasin stake. What it is: ~4.2% post-dilution family stake of unknown encumbrance status, plus Thai authorities actively seizing tycoon assets (SCMP, 5 Jan 2025). Why it matters: even 4% dumped into a low-volume name (60-day median volume ~0.7–1m shares; ฿7.30 spot) is a multi-week overhang. Fair-value discount: 5% liquidity/overhang.

11. What's clean and not a problem (balanced view)

  • The recap itself is a positive forensic event. RAM injected ฿6.28bn cash at ฿8.65 / ฿5.00 in a transparent, SEC-filed transaction (filing THG 21/2025 per dossier). No earn-out, no contingent consideration, no exotic structure — straight equity for cash, used to repay debt. That is clean.
  • Interest savings of >฿300m/yr are a real, mechanical, durable improvement in the P&L — not an accounting choice, just less debt outstanding. This is the highest-quality piece of the FY25 result.
  • Board cleanup is observable: two new independent directors in early 2025, departure of a long-standing one. Consistent with a serious governance reset.
  • No evidence in this package of revenue manipulation, channel stuffing, or aggressive recognition — but absence of evidence isn't evidence of absence; I have no statements to look at.
  • RAM is a credible operator (Ramkhamhaeng Hospital is a long-established listed Thai hospital). A strategic, in-sector buyer is the best possible new owner from an accounting-quality standpoint — they know where the bodies are buried in a hospital P&L.

Bottom line for the LP: Grade D is provisional and driven 70% by the disclosure-blackout and fugitive-founder overhang, 30% by the wafer-thin FY25 earnings on declining revenue. Get the FY25 audited statements (especially the related-party note, contingent liabilities, and post-recap debt schedule) before upgrading or downgrading. Until then, treat any valuation discussion that capitalises FY25 EPS at peer hospital multiples as analytically unsupported.

Word count: ~1,780