Operational Risk Capital Charge: The Standardised Approach

The operational risk capital charge under Basel III’s standardised approach replaces earlier, bank-proprietary advanced models with a formulaic method that combines a bank’s business-indicator component (BIC) with an income-and-losses component (IC). No model approval needed—all banks calculate it the same way—but the result can be just as capital-intensive for banks with large, risky business lines.

The End of Advanced Models

For two decades, large banks built Advanced Measurement Approaches (AMAs) to calculate operational-risk capital. These models were bespoke: a bank would collect historical loss data, calibrate it to its business, and derive a capital charge. The theory was elegant—banks that managed operational-risk better would get lower charges—but implementation bred complexity and inconsistency. Different banks modeled the same risk differently. Model validation consumed armies of risk officers. And during stress periods, when operational losses spiked, banks discovered their models lagged reality.

In 2020, the Basel Committee scrapped AMA and replaced it with a standardised operational risk capital charge. The goal: transparency, comparability, and a formula that captures risk without requiring supervisory approval of proprietary mathematics.

The Two-Component Formula

The standardised approach has two parts: a business indicator and an income component.

Business Indicator (BI)

The BI is a proxy for the bank’s operational risk exposure. It is the sum of net interest income and net non-interest income—essentially, a bank’s gross revenue before provisions and taxes. The logic: a larger, more complex operation faces more operational risk. A bank earning $50 billion annually will likely have greater exposure to operational losses than a $5 billion bank, all else equal.

However, BI alone is crude. A bank might have high revenue but low operational-risk due to tight controls. So the formula adds a margin that penalizes larger banks more, by band:

A $100 billion revenue bank pays 18 percent of BI as its baseline operational risk charge. A $500 million bank pays 12 percent. This graduated structure ensures that systemic risk (concentrated in large banks) attracts proportional capital.

Income Component (IC)

The second pillar, the Income Component, adjusts for actual operational-risk track record. The IC is calculated as the average of operational losses (excluding gains) over a 10-year rolling window, adjusted for inflation and multiplied by a factor (typically between 1.0 and 1.5, set by the Basel Committee). The idea is that a bank with a history of material operational losses should hold more capital, regardless of business size.

Operational losses include:

• Internal fraud: Employee misconduct, rogue trading, misappropriation.
• External fraud: Theft, hacking, payment card fraud, theft of physical assets.
• Employment practices and workplace safety: Discrimination claims, sexual harassment settlements, workplace injuries.
• Clients, products, and business practices: Mis-selling, breach of fiduciary duty, unsuitable advice.
• Damage to physical assets: Disasters, storms, sabotage.
• Business disruption and system failures: IT outages, software bugs, third-party vendor failures.
• Execution, delivery, and process management: Settlement failures, fraud by counterparties, miscalculations, missed deadlines.

Combined Charge

The final operational risk capital charge is the sum of these components. If a bank has a BI of $50 billion and an average 10-year IC of $200 million, and the IC factor is 1.0, the charge might be:

• BIC: $50B × 0.18 = $9.0 billion
• IC: $200M × 1.0 = $200 million
• Total OpRisk charge: $9.2 billion

This charge is converted into a risk-weighted asset equivalent (by dividing by the total capital ratio, typically 8 percent) and added to the bank’s capital-adequacy calculation.

Why This Replaces AMA

The standardised approach offers three advantages over AMA:

Transparency: All banks use the same formula. Regulators can compare operational risk capital across institutions. Investors understand what drives the charge.

Speed: No lengthy model validation cycles. A new large acquirer immediately calculates its charge based on combined BI and IC. Supervisors approve nothing—the math is public.

Procyclicality restraint: AMA models often underestimated risk in booms and overestimated in busts. The standardised formula, because it averages losses over 10 years, smooths this oscillation.

Limitations and Criticisms

The standardised approach is not without flaws. The BI band-based structure is coarse: a $31 billion revenue bank pays 18 percent, while a $29 billion bank pays 15 percent, despite minimal operational risk difference. The IC component only captures realized losses; a bank that barely avoided a major fraud might face a lower charge than one that suffered a minor breach.

Moreover, the 10-year look-back for IC is both generous and harsh. A bank might exit a loss-making business or lose a major source of fraud—but that history lingers, inflating its charge. Conversely, a young bank with no loss history starts with IC = 0, which may not reflect true operational risk.

Interaction with Pillar 1 and Pillar 2

The operational risk charge is part of a bank’s total Pillar-1 minimum capital requirement. It sits alongside credit-risk and market-risk charges. Supervisors, in Pillar 2, may add capital for operational risk concentrations (e.g., a bank with a history of settlement failures) or governance weaknesses.

Some jurisdictions phase in the standardised approach gradually, allowing banks to run it in parallel with legacy AMA models. Others implement immediately, forcing quick recalibration of capital plans.

Data Quality and Granularity

Banks must rigorously classify and measure operational losses to calculate IC accurately. This has driven investment in loss databases and event tracking. Regulators expect banks to defend their IC calculations with detailed loss event documentation. A dispute over whether a $50 million litigation settlement counts as “clients/products” or “employment practices” can shift the IC by millions.

See also