Worth separating the failures here, because the fixes are different.
The regulator's own language is asset quality, not liquidity: solvency fell to 5.56% against a 12% minimum after adjustments during the regularization review, and BCP flagged irregular loans to companies owned by the controlling group. Nobody was queuing at the doors until the resolution was announced. That is a governance failure inside a licensed institution, and no payment rail prevents it. Bitcoin sitting on a custodian's balance sheet would have failed exactly the same way.
The guarantee math is the part worth reading closely. Coverage is 75 minimum wages per depositor, about 228.3M guaranies, and BCP says that reaches 97% of customers by headcount. But the bank administered $800-850M of public funds, and roughly 32,000 retirees had pensions flowing through it. Those balances sit far above the cap, so the fund makes small retail depositors close to whole and converts the pension system into an unsecured creditor of a bank with 5.56% solvency. That recovery is cents on the dollar, years later, and it lands on the same people twice.
The third failure is operational and it is the one anyone building payments should sit with: the interim BCP president said the operational side is the problem in paying out. Advances capped around 10M guaranies, three tranches, emails telling people which institution will pay them. Payout speed is plumbing, and the plumbing ran through the failed entity's own ledger.
Self custody removes the counterparty, but 60% of adults in one app is concentration risk, and if the exit route is a single exchange you rebuilt the same shape with a thinner balance sheet.
Worth separating the failures here, because the fixes are different.
The regulator's own language is asset quality, not liquidity: solvency fell to 5.56% against a 12% minimum after adjustments during the regularization review, and BCP flagged irregular loans to companies owned by the controlling group. Nobody was queuing at the doors until the resolution was announced. That is a governance failure inside a licensed institution, and no payment rail prevents it. Bitcoin sitting on a custodian's balance sheet would have failed exactly the same way.
The guarantee math is the part worth reading closely. Coverage is 75 minimum wages per depositor, about 228.3M guaranies, and BCP says that reaches 97% of customers by headcount. But the bank administered $800-850M of public funds, and roughly 32,000 retirees had pensions flowing through it. Those balances sit far above the cap, so the fund makes small retail depositors close to whole and converts the pension system into an unsecured creditor of a bank with 5.56% solvency. That recovery is cents on the dollar, years later, and it lands on the same people twice.
The third failure is operational and it is the one anyone building payments should sit with: the interim BCP president said the operational side is the problem in paying out. Advances capped around 10M guaranies, three tranches, emails telling people which institution will pay them. Payout speed is plumbing, and the plumbing ran through the failed entity's own ledger.
Self custody removes the counterparty, but 60% of adults in one app is concentration risk, and if the exit route is a single exchange you rebuilt the same shape with a thinner balance sheet.