HongERP — the ESG decision layer

One position · four papers · one operating system. This is an interactive prototype (it requires JavaScript; you are reading the static summary). Source: github.com/JihongParker/hong-erp.

Incumbent ESG software collects disclosures and stops. HongERP treats disclosure as a control variable that lowers the shadow price of residual risk, Λ(d) = φ + λe^−kd, and solves the optimal hedge ratios h* and the optimal disclosure intensity d* as one problem — for a Korean crude-oil importer's joint WTI × USD/KRW exposure, grounded in a four-paper research program by Jihong Park.

Modules (all live)

Decision layer:

• Decision Dashboard — d*, h_f*, h_c* live from a frozen transcription of the ESG-disclosure paper's §3: hedges solve 2Λ(d)Σu = p with KKT corner handling; the voluntary d* solves 2ad = kλe^−kdR; a mandate is a floor d ≥ d̲ (binding floors crowd out hedging). Closed form certified against an independent numerical minimizer on 200 random draws, worst objective gap 3×10⁻⁶.
• Hedge Budget — the budget-allocation paper's convex program verbatim: min σ_res subject to premium + stress ledger ≤ ₩45bn. Reproduces the paper's vertex optimum (97.05% WTI / 2.95% FX coverage, σ 0.0916, budget-exact).
• Hedge Instruments — two tabs. The Vanilla desk is a strategy library refiners actually run — swap, cap, zero-cost collar, three-way collar, and seagull — all priced on Black-76 (put–call parity residual 10⁻¹⁴, the zero-cost structures self-financing to 10⁻¹⁵), plus a crack-spread and FX benchmark. The Quanto desk is the research side: a double knock-out quanto priced on surfaces precomputed from the paper's own jump-diffusion engine (KO probability anchored 43.5% vs the paper's 43.7%), with a Barrier Risk Monitor, a European-quanto ablation, and a textbook lattice (CRR) foil that shows where closed-form deltas break.
• Hedge Accounting — IFRS 9 cash-flow-hedge designation, combined vs split, on the paper engine's 218-step ledgers: mean ineffectiveness ₩23.4bn (combined) vs ₩6.4bn (split), with the split structure's post-knock-out FVTPL noise of ₩71.9bn as the counterweight.

Reporting layer:

• Chart of Accounts / Materiality / Metrics Entry — a sustainability account tree mapped to GRI/KSSB/KCGS/MSCI datapoints, an interactive double-materiality matrix, and a validation-and-approval mockup.

What-if & validation:

• Scenarios — division-level parameters through the same frozen engine.
• Out-of-sample — a walk-forward hedge backtest on 40 years of FRED data (WTI × USD/KRW, 486 monthly returns). The covariance-aware optimum, estimated only on strictly past data and rebalanced monthly after costs, removes ≈89% of out-of-sample cash-flow variance versus ≈75% for a naive even split — this is a hedging result (variance reduction), never alpha or a Sharpe ratio. Includes the paper's dynamic hedge-cost parabola in the coupling c — an exact affine identity whose vertex the paper no longer reads as a covariance-aware hedge ratio: netting the oil futures leg's own FX exposure shifts the vertex by exactly one unit to essentially zero, so the FX coupling is not a risk lever at all (the oil leg dominates the variance). The FX leg is instead sized to the structure's own carrying value, Δ_FX = V/S₂.

The position spine

The modules share one firm-level state: material risks feed the exposure parameters, the budget split lands on the instrument desks, the exotic desk's live knock-out odds drive the accounting module's post-KO exposure, and the disclosure optimum closes the loop. Every screen carries provenance chips saying where each number came from.

The four papers

1. Optimal WTI–FX hedge ratios under a fixed budget (budget allocation)
2. The currency leg of a knock-out quanto: factorization, netting, and the limits of a fitted LSMC delta (delta hedging)
3. IFRS 9 cash-flow-hedge accounting: combined vs split designation
4. ESG disclosure mandates and corporate hedging (Korea / KSSB, staggered DiD)

Values shown are illustrative, never advice.