The Basel Accords

A bank is, at its core, a leveraged bet: it lends out far more than the capital its own shareholders put in, and pockets the spread. That leverage is what makes banking profitable and also what makes a bank's failure so much worse than a normal company's — its creditors are depositors, other banks, and the broader payment system. The Basel Accords are the international answer to a simple question: how much of a bank's own capital has to sit behind its risks before regulators consider it safe enough to operate?

Why "risk-weighted"

The core idea, present since Basel I (1988), is that a bank doesn't hold capital against its raw assets, it holds capital against risk-weighted assets. A dollar of loan to a government is treated as close to riskless and needs little capital behind it; a dollar of unsecured lending to a risky corporate needs much more. The headline requirement is a ratio:

$\text{Capital ratio} = \frac{\text{Bank's own capital}}{\text{Risk-weighted assets}} \geq \text{minimum \%}.$

In words: take the bank's loss-absorbing capital, divide by a measure of how risky its book is (not its raw size), and that ratio must clear a floor. A bank can grow its balance sheet as much as it wants, as long as it keeps raising or retaining enough capital to keep the ratio above the line.

The three versions, and what broke each one

Basel I used a handful of broad categories, government debt at 0% weight, mortgages at 50%, most corporate loans at 100%, regardless of the actual borrower's quality. This was simple but crude: a AAA corporate and a near-junk corporate needed the same capital, which pushed banks toward the riskier end of each bucket to earn more return for the same regulatory cost.

Basel II let large banks use their own internal models to estimate risk weights, more accurate in principle, but it let banks that wanted low capital requirements build models that produced them. In the run-up to 2008, risk weights on AAA-rated mortgage securitizations were extremely low, right up until those securities turned out to be far riskier than modeled. The framework didn't require enough capital against exactly the risks that blew up.

Basel III, written after 2008, raised both the quantity and quality of required capital (more common equity, less of the hybrid instruments that had proven not to absorb losses in practice), added a simple leverage ratio that ignores risk-weighting entirely as a backstop against gamed models, and introduced liquidity requirements — the Liquidity Coverage Ratio and Net Stable Funding Ratio — that address Funding Liquidity Risk directly, something earlier accords never touched.

Worked example

A bank holds $100m in AAA-rated sovereign bonds (0% risk weight) and $100m in unsecured corporate loans (100% risk weight). Risk-weighted assets are $0.00 \times 100\text{m} + 1.00 \times 100\text{m} =$ $100m, not $200m — the sovereign bonds contribute nothing to the requirement. Under an 8% minimum ratio, the bank needs $0.08 \times 100\text{m} =$ $8m of capital. If it instead shifted the sovereign holding into the corporate book, RWA would double to $200m and the capital requirement would double to $16m for the same total balance sheet size — which is exactly the incentive risk-weighting is designed to create: hold more capital only where you're taking more risk.

What this means in practice

• A bank's capital ratio is the single number regulators and analysts watch most closely as a health signal, alongside the leverage ratio as a sanity check.
• FRTB, covered in FRTB: The Fundamental Review of the Trading Book, is Basel III's specific rewrite of how trading book risk gets measured, tightening exactly the kind of internal-model gaming that Basel II allowed.
• Counterparty Credit Risk capital charges and margin rules for derivatives were both substantially rewritten under Basel III, for the same reason — 2008 revealed both were previously underpriced.