Braskem restructuring - negotiated outcome analysis

Public information through 2026-08-15. Four independent specialists, three rounds, then neutral mediation.

Post-run correction, 16 August 2026. A targeted Braskem, creditor and neutral-mediator review corrected the treatment of Q2 working capital. It also corrected the H2 operational working-capital release in the June plan to US$518M, not US$548M. The probability refresh below reflects those corrections. The structure is unchanged, but the revised clearing estimate uses 15% upfront creditor warrants, stepping to 22.5% if cash-conversion, liquidity or PIK milestones are missed.

Working-capital correction

Persistent spreads can improve cash generation if naphtha prices and inventory volumes stabilize. Q2's US$547M outflow included higher feedstock and product prices, higher inventory volumes, and reduced payment arrangements. The price and inventory components are balance-sheet level changes that should not repeat without another increase, but Braskem did not disclose the exact split and the cash remains tied up until balances reverse. The financing component is more persistent.

Q2 therefore demonstrates a stressed historical bridge, not a structural 20% conversion rate: US$1,043M EBITDA → US$210M recurring cash → negative US$149M of actual cash movement. Adding back the full working-capital build while holding every other bridge item constant produces approximately US$757M of recurring cash. That figure is an illustrative sensitivity, not normalized cash flow or a forecast. Management's plan assumes a +US$518M H2-26 operational working-capital release. A defensible sensitivity range is flat operational working capital to that full release, rather than mechanically repeating another Q2-sized outflow.

Spreads can improve both the price of the deal and near-term cash generation, but they do not remove the need for a restructuring. The remaining structural constraints are trade finance and maturities: the LC book runs 1,274 → 502 → zero in 2027, reverse factoring is gone, supplier advances tripled, and R$1,632M of LCs have already been drawn by banks and reclassified as loans.

Post-run Hormuz scenario, 16 August 2026

If Hormuz remains effectively closed and PE minus naphtha spreads stay near current strong levels, the conditional central estimate is US$2.7B of annual EBITDA, with a US$2.5B to US$2.9B range. The year estimates are US$2.8B to US$3.1B for 2026, US$2.6B to US$3.0B for 2027 and US$2.5B to US$2.9B for 2028. At US$2.7B of sustainable end-2028 EBITDA, the equity-option model implies BAK at US$11.14, with an US$8.15 to US$14.20 range.

Because PE minus naphtha is a spread measure, higher naphtha cost is already reflected in the EBITDA estimate. Higher feedstock and product prices can still create a working-capital build through more expensive inventory and receivables. If prices and inventory stabilize, that balance-driven use should not repeat at the same scale, although the exact temporary portion of Q2's outflow was not disclosed.

This is a conditional scenario, not the probability-weighted base case. Alternative trade routes, higher Chinese coal-to-olefins output and a partial normalization of Strait traffic can erode the supply shock. Weighting those paths lowers the all-scenarios estimate to roughly US$2.3B to US$2.5B. The forecast and BAK values are model inferences, not management guidance.

Spread-to-EBITDA sensitivity

The international PE minus naphtha benchmark and Braskem's effective domestic resin spreads are different measures. Used together, they support the following directional map for sustainable annual EBITDA:

Calibration uses Q2 2026 actual spreads of US$773/t for the benchmark, US$1,738/t for effective domestic PE and US$622/t for effective domestic PP, alongside US$1,043M of quarterly EBITDA. The June plan modeled effective PE spreads of US$1,088/t in Q3 and US$977/t in Q4, with quarterly EBITDA of US$586M and US$351M. The sensitivity uses approximately US$200M of annual EBITDA for each sustained US$100/t change in the PE benchmark, then adjusts directionally for PP, chemicals, international businesses, utilization and REIQ. It is nonlinear scenario analysis, not company guidance or investment advice.

A. Most likely negotiated deal - 40%

1. Why creditors accept instead of forcing RJ

Their own arithmetic condemns the threat: bonds mark ~58c, RJ returns ~35c after a 15–25% process discount and 2–4 years, the consensual package is worth 65–73c. They settle ~30 points above their BATNA. The hostility is about capturing the top of the zone, not preferring RJ. Their credible threat rests on asymmetry, not enthusiasm - "we lose 23 points; they lose 100%."

2. Why Petrobras contributes instead of walking away

US$750M protects a 36.15% economic stake, joint control, and an integrated Camaçari/Rio feedstock relationship. Crucially, only ~US$500M is a cheque; the rest is 60-day commercial payment terms requiring no capital approval, no independent laudo, no TCU-exposed related-party event. That is what makes it deliverable seven weeks before an election. And Petrobras stays at 47.03% voting - crossing 50% would consolidate ~US$10B of debt onto its balance sheet.

3. Why IG4 accepts dilution

The most surprising output of the exercise: at US$1.5B EBITDA, IG4 is strictly better off being crammed down (US$275–315M) than winning a clean no-haircut extension (−US$154M net of its cheque). Deleveraging beats percentage. Its real red line is therefore not "no equitization" - it is no equitization at a trough price and none I'm not paid an option for. It demands a contingent equitization rather than refusing one, buying a levered option on ≥US$2.0B EBITDA.

4. Why Braskem can service the structure

Cash interest falls from ~US$714M to ~US$430M for 24 months; closing liquidity ~US$2.2B plus US$1.2B of LC capacity. Net leverage at close: 5.22x at US$1.5B EBITDA, 3.92x at US$2.0B, 3.13x at US$2.5B, 2.61x at US$3.0B, 1.96x at US$4.0B. At US$1.5B it does not work - which is exactly what the warrant ladder and the Dec-2028 contingent equitization are for. That contingency is not a sweetener; it is load-bearing.

Indicative end-2028 BAK valuation sensitivity. A simple residual-value calculation incorrectly assigns zero to equity whenever enterprise value is below debt, even though the maturity extension preserves option value. The table therefore treats equity as a call option on enterprise value. The midpoint uses 5.0x EV/EBITDA, US$10.26B net debt after modeled PIK, three years to extended maturity, 30% asset volatility, a 4.5% risk-free rate, two shares per ADR, equal economic value across share classes and 15% warrant dilution.

Each EBITDA case is an annualized run rate at end-2028, not cumulative EBITDA over two years. The range varies the multiple from 4.5x to 5.5x and asset volatility from 25% to 35%. It excludes intervening free cash flow, further cash burn and the 22.5% warrant step-up. This is scenario analysis, not a price target.

5. What existing BAK/BRKM5 holders retain

The warrant grant alone leaves every legacy holder with 85% of their former ownership percentage and requires no cash contribution. If the sponsor funding is implemented through the modeled rights offering, minorities hold approximately 25.1% if they subscribe and 12.4% if they do not. See G–I.

6. The strongest argument against this conclusion

Revealed behaviour over the last 60 days still argues for caution: four failed negotiating rounds, disagreement over liens, and a controller that let a Valor report stand unrebutted saying RJ already has its unofficial consent. The equity raise also cannot close before the stay lapses (~25 Aug) or before the election. The working-capital correction modestly improves the consensual path because a strong-spread quarter can generate cash once balances stabilize. It does not eliminate execution risk or the maturity wall, which is why failure/RJ remains a substantial 28% outcome.

The lien crux, and why it clears

The dispute is not about value, it is about states of the world. Creditors want the lien for the normalization state; the company needs assets unencumbered in the persistent state because unencumbered assets are what support the LC and factoring lines - its actual binding constraint. Granting liens at close is RJ damage without RJ benefits.

A pre-signed, pre-perfected second lien over the Brazilian crackers that springs at net leverage >4.5x costs the company nothing in the state where it doesn't spring and delivers creditors near-full protection in the state where it does. Petrobras and IG4 proposed variants independently. It clears - and the same logic drives the warrant ladder, which prices the spread disagreement instead of requiring the parties to resolve it.

B–D. The other outcomes

Outcome C is consensual only because IG4 would rationally vote for it at low EBITDA - per its own BATNA, deleveraging beats percentage. Absent that vote it is legally unreachable (see G).

E–F. Dilution under each

Base 797.2M shares (451.7M ON / 345.1M PNA). The base warrant grant alone leaves legacy shareholders at 85% fully diluted. The table below also illustrates a separate US$1.1B rights offering at ~R$7.00: 818.7M new shares → 1,615.9M before warrants; 15% warrants → approximately 1,901.0M fully diluted.

Under the rights-offering illustration, a non-subscribing minority loses about 58% of its ownership share; a full subscriber loses only the 15% warrant dilution. Percentage dilution is not value destruction - a subscriber pays R$7.00 for shares worth materially more at US$2.0B+ EBITDA. A non-subscriber suffers no value transfer only if the issue price is fair, which is why pricing near the R$6.36 market, not at a trough, is the fairness anchor.

G. Do IG4/Petrobras retain control?

Yes, in A and B. The rights offering issues in current class proportion, so if both sponsors subscribe, Shine holds 50.11% of ON and Petrobras 47.03% throughout. Warrants settle into non-voting preferred - mandatory, because Petrobras cannot approach 50% without consolidating ~US$10B of debt, and IG4 will not go below ~35% of ON.

The decisive legal fact: authorized capital headroom is only 355.7M of 1,152,937,970 shares - ~30.85% of a post-money. Anything larger requires an EGM, which requires an ON majority, which only Shine can deliver. Consensual equitization above ~31% is legally impossible without IG4's affirmative vote. Creditors who want more equity must go through RJ. This is IG4's hardest lever, and it helps explain why outcome C is 15% while D is 28%.

H–I. Can BAK/BRKM5 holders participate, and what if they don't?

BRKM5 holders: yes. Art. 171 §1(b) gives preemption within class first, then cross-class for leftovers; §6 makes rights freely assignable. Expect a ≥30-day window with rights listed on B3 plus a sobras round. PNA retains its priority non-cumulative 6% dividend and Class A's equal claim on residual profit; new PNA must carry identical rights.

BAK/ADS holders (13.24% of capital, 30.6% of the PNA class): no cash contribution is required by the base warrant package. If a later cash rights offering is used, direct ADS participation depends on the filed US tranche and depositary procedure. The depositary may facilitate exercise, distribute or sell rights, or allow them to lapse depending on legality and practicality. Do not assume direct participation or forced exclusion before transaction documents exist.

Most actionable finding for a BAK holder: do not convert solely because of the base warrant grant; BAK and BRKM5 are diluted equally. Reassess only if Braskem files a cash rights offering whose ADS procedures differ from direct BRKM5 participation.

If a later rights offering occurs and minorities do not subscribe: sponsors backstop the shortfall in class-A preferred only - capping voting drift and blocking an art. 117 abuse-of-control claim. Under the illustration, minorities fall from 29.53% to ~14.6% before warrants and ~12.4% at 15% warrants, while Petrobras and Shine increase economically without gaining a vote.

J. Events that move these probabilities

Near term, highest signal:

1. ~25 Aug - stay expiry. EJ filing → A/B/C. Lapse with no filing → D sharply. The board's own runway analysis says no-filing is cash-negative in 45–75 days.
2. Any lock-up or RSA announcement before ~30 Sep - the political close window. Slippage pushes execution into Q1-27, across the Dec-26 RCF wall, and raises D materially.
3. Whether Petrobras delivers the 60-day payment terms - needs no board capital approval, so it is the cleanest early tell of genuine sponsor support.

Then:

4. 4 Oct / 25 Oct elections - Petrobras's cheque is effectively frozen between late Sept and Q1-27.
5. 16 Oct - 20% resin import tariff expires (GECEX 800/2025), between the election rounds. Non-renewal is worth roughly the gap between 70% and 75%+ Brazilian utilization and triggers a warrant step-up in every proposed structure.
6. 31 Dec - RCF/stand-by US$1,004M matures. Hard wall.
7. 3Q26 results (Nov) - first hard evidence on whether spreads held and converted to cash. The single largest probability-mover between A and C.
8. Braskem Idesa Chapter 11 filing - if the perimeter leaks into Braskem S.A., all bets reprice.
9. Whether the AHG grants deep-dive diligence on projections - creditors listed refusal as a walk-away condition.
10. Any BNDES facility - would substitute for sponsor cash; the most plausible official-sector circuit-breaker in an election year.

Method note

All four openings were formed independently - no agent saw another's role file or position before writing. The convergence (zero haircut, no '41/'50 extension, coupon increase, Mexico ring-fence, Alagoas carve-out, ~24-month relief at a 4% cash floor) emerged without coordination. A targeted working-capital refresh was then performed by the Braskem, creditor and neutral-mediator reviewers. That refresh raises A+B+C to 72% while the negotiating record and July 29 non-denial keep D at 28%.