Quick answer: economic capital is an internal estimate of how much loss-absorbing capacity a bank needs for the risks it chooses to take, measured under its own risk framework rather than simply copied from a regulatory formula. Once the bank has a total economic-capital number, it still has a second problem: which business line, portfolio or transaction should be charged for how much of it? Stand-alone capital ignores diversification. Marginal allocation asks how total risk changes if an exposure changes slightly. Euler allocation turns that marginal effect into additive risk contributions that sum back to the portfolio total when the risk measure has the required mathematical properties. Those allocations can then feed RAROC, pricing, limits and risk budgets.

Capital allocation is not just dividing a pie. The size of each slice depends on how each business changes the risk of the whole bank.

Ownership boundary: total capital is not the same job as allocating it

The Bukit Timah Tutor article How Bank Capital Models Turn Risk Into Constraints explains regulatory capital ratios, RWA and expected versus unexpected loss. The newer balance-sheet optimisation article asks how the bank chooses activities subject to several constraints.

This page owns the narrower mathematical problem in between:

If the bank decides that its portfolio requires X units of internal risk capital, how should X be attributed to the businesses that created the risk?

1. Economic capital is an internal risk measure

The Basel Committee describes economic-capital frameworks as methods that allow banks to assess risk consistently and attribute capital to cover the economic effects of risk-taking activities. Economic capital can feed internal capital adequacy, pricing, portfolio management, performance measurement and incentives, but it remains subject to major modelling and governance challenges.

See Basel Committee — Range of practices and issues in economic capital frameworks.

A stylised economic-capital definition is:

Economic Capital = high-quantile portfolio loss − expected loss

over a defined horizon and confidence standard. Some frameworks use other risk measures, including Expected Shortfall or internally defined stress measures. The precise metric is a governance choice that must match the risk and intended use.

2. Why stand-alone capital overstates the sum

Suppose Business A requires S$100 million of economic capital if it existed alone, and Business B requires S$80 million alone. If their losses are imperfectly correlated, the combined bank may need only S$150 million rather than S$180 million.

The difference—S$30 million—is a diversification benefit created by combining risks that do not peak in exactly the same states.

This immediately creates an allocation problem. If A is charged 100 and B is charged 80, business-line capital sums to more than the bank actually needs. If management simply subtracts 15 from each, the result is easy but not necessarily connected to how each business creates diversification.

3. Marginal risk: what happens if this business grows slightly?

Let the bank’s total risk measure be ρ(x), where x is the vector of business exposures. The marginal risk contribution of business i is the derivative:

MRC_i = ∂ρ / ∂x_i.

This asks a local question: if the bank increased exposure to business i a tiny amount while everything else stayed approximately fixed, how much would total portfolio risk rise?

A business can have high stand-alone volatility but low or even negative marginal contribution if it offsets risk already concentrated elsewhere. Conversely, a seemingly moderate business can have high marginal contribution if it loads heavily on the bank’s dominant stress factor.

4. Euler allocation makes the contributions add back to total risk

For a differentiable, positively homogeneous risk measure, Euler’s theorem gives a particularly useful allocation:

RC_i = x_i × ∂ρ/∂x_i.

Then:

Σ RC_i = ρ(x).

This property is powerful because the allocated risk contributions sum exactly to the portfolio risk measure instead of creating an unexplained reconciliation difference.

The allocation is also economically meaningful: a business is charged according to both its size and how its next unit of exposure changes the whole portfolio.

5. A two-business example with correlation

Suppose two business lines have stand-alone risk measures σ_A=100 and σ_B=80, with correlation ρ=0.25. A simplified variance-style portfolio risk is:

R = √(100² + 80² + 2×0.25×100×80).

This produces approximately 142.8 units of total risk—not 180.

Under an Euler decomposition of this volatility-style measure, Business A contributes roughly:

(100² + 0.25×100×80) / 142.8 ≈ 84.0

and Business B contributes roughly 58.8. The contributions sum to 142.8.

The allocation is not proportional to stand-alone risk because each line receives a share of the diversification effect through its covariance with the other.

6. Diversification is a model output, not free money

If the correlation assumption changes from 0.25 to 0.80 in stress, combined risk increases sharply. A business line whose apparent capital efficiency depends on diversification can become much more expensive when common factors strengthen.

This is why Basel’s economic-capital review warned that risk aggregation and dependency modelling are among the most difficult parts of internal capital frameworks. Correlation is not a permanent physical constant. It is an estimate that can change precisely when losses become large.

The same lesson appears in credit-portfolio concentration: many exposures are not diversified if they depend on the same underlying state.

7. Allocation across risk types is harder than allocation within one portfolio

A bank may estimate economic capital for credit risk, market risk, operational risk, IRRBB and other material risks. Summing them assumes zero diversification. Applying one giant correlation matrix assumes the dependencies are measurable and stable.

Cross-risk aggregation therefore often uses combinations of:

variance-covariance structures;
copulas or simulation;
scenario aggregation;
conservative floors;
partial recognition of diversification;
management overlays where evidence is insufficient.

The allocation method should be no more confident than the dependency evidence underneath it.

8. RAROC turns allocated capital into a performance denominator

Once economic capital is allocated, a bank can calculate a risk-adjusted performance ratio such as:

RAROC = risk-adjusted income / allocated economic capital.

Suppose Business A earns S$20 million of risk-adjusted income and receives S$100 million of allocated capital: RAROC=20%. Business B earns S$18 million on S$150 million of allocated capital: RAROC=12%.

The higher raw profit of one line can therefore be less attractive after capital intensity is recognised. But RAROC is only as reliable as the capital allocation. Understate a business’s tail correlation and its apparent performance improves without any real economic improvement.

For transaction-level pricing, see How Banks Price Loans for Risk-Adjusted Return.

9. Risk budgets turn allocation into a constraint

Instead of using allocated capital only for reporting, management can assign risk budgets. A business may be permitted a maximum share of total economic capital, marginal tail risk or stress loss.

A budget can be written:

RC_i ≤ Budget_i.

If a business exceeds its risk budget, it can reduce exposure, hedge, improve collateral, raise pricing, sell assets or request more capital allocation. This turns an abstract risk measure into a resource-allocation rule.

10. Marginal RAROC helps decide whether the next deal improves the bank

Average RAROC asks how an existing business has performed. Marginal RAROC asks whether adding a new transaction improves or worsens portfolio value.

A new loan can have modest stand-alone profitability but be attractive because it diversifies the portfolio. Another loan can look profitable alone but be unattractive because it loads onto an already concentrated sector.

This is the connection between economic-capital allocation and portfolio optimisation: the bank should care about the incremental effect of the next decision, not only the average performance of the existing book.

11. Capital transfer pricing can change business behaviour

If internal management charges 12% per year for economic capital, a business using S$100 million of capital receives a S$12 million capital charge in its profitability view.

The charge encourages businesses to ask:

Can the same revenue be produced with less tail risk?
Can collateral reduce loss severity?
Can hedging reduce market or basis exposure?
Should a concentrated portfolio be sold or syndicated?
Does pricing cover the capital resource consumed?

A bad allocation can train the organisation in the wrong direction. If diversification credit is given too generously, correlated businesses can appear cheap. If diversification is ignored completely, genuinely offsetting businesses can look expensive and be discouraged.

12. Creative-work lens: Master and Commander and limited fighting capacity

Master and Commander is not a banking source, but it offers a useful allocation lens. A ship has limited crew, powder, repair capacity and attention. Assigning every resource to the strongest gun position can leave sails, medicine or damage control dangerously under-resourced. The value of one allocation depends on what the whole system already contains.

Economic capital works similarly: a business line cannot be judged only in isolation. Its value and cost depend on how its loss states overlap with the rest of the bank. The creative work provides the intuition; the allocation still needs quantitative risk contributions and evidence.

13. The allocation pipeline

Define the internal risk measure and horizon.
Estimate stand-alone risk for each portfolio or business.
Model dependencies and aggregate to total economic capital.
Calculate marginal contributions to portfolio risk.
Use Euler or another governed allocation method where appropriate.
Reconcile allocated contributions back to the total.
Apply stress scenarios that challenge diversification assumptions.
Calculate RAROC or economic profit using allocated capital.
Set risk budgets and limits.
Use marginal analysis for new transactions and growth plans.
Track business responses to the capital charge.
Backtest whether high-capital businesses actually generate the tail losses assumed.
Recalibrate dependencies when regimes or portfolios change.

14. Failure modes

Stand-alone summation. Every business is charged as if diversification does not exist.
Diversification fantasy. Correlation benefits are assumed stable in stress.
Non-additive allocation. Business charges do not reconcile to total capital.
Average-profit bias. High absolute profit hides poor return on scarce risk capacity.
Hurdle-rate gaming. Capital methodology is tuned until favoured businesses clear the target.
Cross-risk overconfidence. Credit, market and operational risks are aggregated with weak dependency evidence.
Historical-normality bias. Capital is calibrated to a period that excludes the bank’s real tail risk.
Metric-only management. Business decisions optimise RAROC while ignoring liquidity, strategic or regulatory constraints.

15. Diagnostics and falsifiers

Do allocated contributions sum to total economic capital?
Which business receives the largest diversification benefit?
What happens to that benefit if correlations move toward stressed values?
Which business has low average RAROC but improves portfolio diversification?
Which business has high stand-alone RAROC but high marginal tail contribution?
How sensitive are risk budgets to the choice of VaR versus Expected Shortfall or stress measure?
Can management explain the allocation change from last quarter?
What observation would show that the dependency model is too optimistic?

Suppose someone claims, “Business A is efficient because its stand-alone capital is low.” A falsifier is evidence that A is highly correlated with the bank’s dominant loss factor and therefore has a large marginal contribution to portfolio risk. Stand-alone risk and portfolio contribution are different objects.

16. Verification and update triggers

reconcile allocation to the approved total economic-capital measure;
independently reproduce marginal/Euler calculations;
stress dependency assumptions and diversification credits;
compare allocation outcomes with realised stress losses;
review whether capital charges create perverse business incentives;
update after acquisitions, portfolio sales or material concentration changes;
keep regulatory capital and economic capital definitions separate;
retain a simpler benchmark allocation to challenge the production method.

Connections across the finance-and-banking algorithms lane

Regulatory capital models — a separate prudential framework from internal economic capital.
Risk-adjusted loan pricing — transaction-level use of allocated capital and hurdle returns.
Credit concentration — a major source of marginal capital contribution.
Balance-sheet optimisation — uses scarce-capital constraints after the capital resource has been measured and allocated.

Research anchors

The deeper lesson

Economic-capital allocation is the mathematics of shared risk. Stand-alone measures ask what a business would cost by itself. Marginal contributions ask what the business adds to the actual bank. Euler allocation makes those contributions additive. RAROC turns them into performance measures. Risk budgets turn them into constraints. The strongest allocation system therefore does not reward a business for looking safe alone; it asks how that business changes the risk of the whole institution.

Educational note: This article explains public banking and risk-allocation concepts. It is not capital-planning advice, investment advice or an internal-capital model for any institution.

Small Group Tutorials

How Banks Allocate Economic Capital Across Business Lines: Euler Allocation, Marginal Risk, Diversification, RAROC and Risk Budgets