Skip to content
Learning path Gold Professional operator

Risk budgeting

A method that assigns target contributions to a risk measure and builds or controls a portfolio so that observed contributions remain consistent with those budgets.

Who this is for — Anyone who wants to state in advance how much each component should participate in a selected risk measure and then compare that objective with measured contributions.

A risk budget is a target assigned to a portfolio component. Risk budgeting is the process that translates those targets into weights or limits and monitors their deviations. The budget is not observed risk: measuring the latter requires a risk contribution. Nor is a risk budget invested capital, notional exposure or margin.

The word “risk” remains incomplete without a metric. A budget may refer to volatility, Expected Shortfall, economic capital or scenario loss. Change the measure and the contributions, weights and interpretation all change. A portfolio balanced by volatility may still be concentrated in liquidity, default or tail risk.

Capital and risk: connected ledgers, not interchangeable ones Architecture starts with resources and ends in traceable limits, measures and decisions Capital and risk: connected ledgers, not interchangeable ones Architecture starts with resources and ends in traceable limits, measures and decisions 1 · Available capital Resources, liquidity,collateral and constraintsdefine what can be deployed. 2 · Allocation Weights or amounts distributecapital across strategies,instruments and reserve. 3 · Risk budget Ex-ante targets allocateshares of a measure under astated horizon and perimeter. 4 · Measure and control Contributions, concentrationsand stress show whether actualuse fits the architecture. Capital funds exposures; a risk budget assigns target shares of ameasure. Equal money does not imply equal risk. Cyclepedia · source-checked conceptual map
The budget belongs to the decision; the contribution belongs to the portfolio and model; the weight is one of the levers available to bring them closer.

Relative and absolute budgets

Let R(w) be a risk measure and RCᵢ(w) the contribution of component i. A vector of relative budgets satisfies:

bᵢ ≥ 0;   ∑ᵢ bᵢ = 1

The objective condition in a risk-budgeting portfolio is:

RCᵢ(w) / R(w) = bᵢ

or, equivalently when total risk is positive:

RCᵢ(w) = bᵢ × R(w)

A relative budget divides total risk among components. An absolute budget adds a desired level R*, such as a portfolio-volatility target, and assigns bᵢ × R*. Relative distribution and total scale are separate decisions: one can first identify weights that meet the shares and then scale exposure within leverage, cash, funding and other limits. Volatility targeting is one possible scaling mechanism, not an automatic consequence of risk budgeting.

Full allocation, ∑ RCᵢ = R, follows from the Euler principle when the measure is differentiable and homogeneous of degree one. If those conditions fail, the methodology must explain how any residual is attributed and why the selected budgets can be compared with the calculated contributions.


Four objects that must remain separate

Object Nature Example question
Capital weight assigned quantity what share of value is in strategy A?
Risk contribution model output for the current portfolio how much volatility is attributed to A?
Risk budget allocation target what share of volatility should A receive?
Risk limit constraint or escalation threshold what value must A not exceed under the mandate?

A budget need not be reached at any cost. A limit can override a target; a position may not be tradable in its theoretical quantity; and a budget can be incompatible with concentration, liquidity or regulation. A deviation does not necessarily trigger a trade. First attribute it to prices, estimates, cash flows, rounding or an intentional mandate change.


From mandate to weights

A complete process includes at least the following steps:

  1. Choose the scope and hierarchy. Budgets can be assigned to asset classes, strategies, factors or desks. Hierarchies must avoid counting the same exposure twice.
  2. Define the measure. Formula, horizon, data, currency and treatment of nonlinear payoffs must be documented.
  3. Set the budgets. They may follow from the mandate, loss capacity or a strategic decision; they are not estimated from price history alone.
  4. Impose real constraints. Long-only rules, leverage, turnover, concentration, liquidity, collateral and tradable lots bound the possible solutions.
  5. Solve and verify. The numerical method seeks weights whose contributions match the budgets; convergence and residuals need checking.
  6. Stress the solution. Alternative matrices, shocks, costs and exit capacity reveal whether the balance is fragile.
  7. Implement and monitor. Actual weights and contributions move; bands, frequency and escalation depend on the mandate and costs.

Existence and uniqueness depend on the measure, matrix, permitted signs and constraints. Results known for long-only volatility portfolios with a well-behaved matrix must not be extended automatically to short positions, derivatives, zero budgets or non-smooth measures.


Example: budgets are not weights

Two uncorrelated assets have estimated volatilities of 10% and 20%. Desired contribution shares to portfolio volatility are 75% and 25%. In the uncorrelated case, each contribution share is proportional to (wᵢ × σᵢ)². Meeting the budgets therefore requires:

w₁ × 0.10 : w₂ × 0.20 = √0.75 : √0.25

With normalized long-only weights, the ratio gives about 77.6% in the first asset and 22.4% in the second. Those figures are not 75/25: the second asset needs less capital because its standalone volatility is higher. If correlation were not zero, the solution would change because every contribution would also include covariance.

The example illustrates a mathematical relationship under simplified assumptions. It is not a recommended allocation and does not include return, tail risk, cost, liquidity or estimation risk.


Risk parity as a special case

Risk parity, in the equal risk contribution sense, sets bᵢ = 1/n. It is a special case of risk budgeting, not a synonym for the entire process. Unequal budgets can express a strategic mandate; hierarchical budgets may first divide risk across asset classes and then among their internal components.

Inverse-volatility weights are a shortcut that coincides with equal contributions only under restrictive dependence structures. A full risk-budgeting calculation uses the chosen joint matrix or model. Even when budgets are met exactly ex ante, new data can move contributions ex post.


Model risk and governance

Precise-looking budgets can create an illusion of control when inputs are unstable. Covariances, volatilities and tail returns change across samples and regimes. Illiquid components can display artificially low volatility because of stale pricing and consequently receive excessive theoretical weight. Strategies with embedded leverage or optionality require look-through and coherent revaluation.

A sound control process retains the model version, data, approved budgets, constraints, deviations and decisions. Estimation frequency, bands and caps are not universal; they must reflect horizon, liquidity, turnover and capacity to act. It is useful to publish target, estimated, stressed and realized contributions together without assigning certainty to their comparison.

Common mistake — Writing “10% risk budget” without saying 10% of which risk, of what total and under which model. A percentage without measure, horizon and scope is not verifiable.


Sources