Proposta de metodologia simplificada para o cálculo dos indicadores previstos nas novas regras do FGC[Proposal of a simplified methodology for the calculation of indicators foreseen in the new FGC rules]

The 2026 amendments to Brazil's deposit insurance framework — administered by the Fundo Garantidor de Créditos (FGC) — introduced through CMN Resolution No. 5,295/26 and BCB Resolution No. 572/26, created the Reference Asset (Ativo de Referência, AR) as a third, independent trigger for the mandatory allocation of federal government bonds (MATPF), alongside existing thresholds based on the ratio between the Reference Value (VR) and Adjusted Net Equity (PLA). Computing the AR through the official route requires contract-level asset-to-liability matching, economic hedge treatment, and fund look-through granularity rarely available to small and mid-sized institutions. This paper draws on minutes from a meeting between the Central Bank of Brazil and the Brazilian Banking Association to show why proportional allocation is both regulatorily prohibited and methodologically unsound, and proposes an alternative named-deduction method: aggregating the asset blocks admitted under the rule, then subtracting only the items it expressly names, with a closing test against total balance-sheet assets. The methodology is applied to a reference case computed from internal data and tested on five additional institutions of similar profile — wholesale banks serving corporate clients, with credit as their core product — using only publicly available audited financial statements. Results show that proportional allocation produces an unpredictable sign error, ranging from −29.4% to +60.6% relative to the named-deduction method, distorting the regulatory buffer in both directions. The named-deduction approach, by contrast, proves auditable, low-cost to implement, and applicable directly from published financial statements — enabling third parties, including investors, treasury desks, and credit departments, to estimate these indicators for institutions that do not disclose them.