How does risk budgeting work in portfolio construction and what are its practical applications?
FRM II discusses risk budgeting as an alternative to traditional return-based portfolio construction. I understand it allocates risk (not capital) across strategies or asset classes, but how do you actually implement it? And how is it different from just targeting a VaR limit?
Risk budgeting is a portfolio construction framework that allocates a total risk budget across portfolio components, ensuring each position or strategy contributes the intended amount of risk. Unlike traditional approaches that start with return targets, risk budgeting starts with risk and asks: 'How should we distribute our risk-taking capacity?'
Why Risk Budgeting, Not Return Targeting?
Return estimates are notoriously unreliable (expected return estimates have wide confidence intervals). Risk estimates (volatility, correlation) are more stable and predictable. By focusing on risk allocation, you build portfolios that are robust to return estimation errors.
The Framework:
Total Risk Budget = The maximum amount of risk the portfolio can take (expressed as VaR, tracking error, or volatility)
Risk Budget Allocation = How the total budget is distributed across components
Key Concept: Marginal Contribution to Risk (MCTR) and Component Contribution to Risk (CCTR)
- MCTR of asset i = How much total portfolio risk changes if we increase the weight of asset i by a small amount
- CCTR of asset i = Weight of asset i × MCTR_i = The portion of total portfolio risk attributable to asset i
- Sum of all CCTR_i = Total portfolio risk
Example:
A multi-strategy fund at Blackridge Partners has a total VaR budget of $50 million:
| Strategy | Capital Allocation | Risk Budget | Actual VaR | Status |
|---|---|---|---|---|
| Global Macro | 30% ($300M) | $20M VaR | $18M | Under budget |
| Credit Long/Short | 25% ($250M) | $15M VaR | $16M | Slightly over |
| Equity Market Neutral | 25% ($250M) | $8M VaR | $7M | Under budget |
| Systematic CTA | 20% ($200M) | $7M VaR | $9M | Over budget |
| Total | 100% | $50M | $50M | — |
Note that capital allocation (30/25/25/20) differs significantly from risk allocation (40/30/16/14%). Global Macro takes 30% of capital but 40% of risk, while Equity Market Neutral takes 25% of capital but only 16% of risk.
Risk Parity — A Special Case:
Risk parity is a risk budgeting approach where each asset class contributes equally to total portfolio risk. This typically results in:
- Higher allocation to bonds (low volatility → needs more capital to generate equal risk contribution)
- Lower allocation to equities (high volatility → less capital needed)
- Often uses leverage to achieve target return level
Practical Applications:
- Multi-strategy hedge funds — Allocate VaR budgets across trading desks
- Asset owners — Distribute risk across asset classes rather than using fixed capital weights
- Active management — Budget tracking error across sectors or portfolio managers
- Risk limits enforcement — Dynamic rebalancing when a component exceeds its risk budget
Difference from Simple VaR Limits:
A VaR limit says 'don't exceed $50M VaR total.' Risk budgeting says 'your $50M must be distributed with $20M to macro, $15M to credit, etc.' The budget ensures intentional risk allocation rather than letting the most aggressive desk consume the entire budget.
Exam Tip: FRM II tests the concepts of MCTR, CCTR, risk parity, and the rationale for focusing on risk rather than return in portfolio construction.
Build your risk budgeting skills in our FRM Part II practice questions.
Master Part II with our FRM Course
64 lessons · 120+ hours· Expert instruction
Related Questions
Why is DV01 so much smaller than dollar duration if both are supposed to measure rate risk?
When should I stop using modified duration and switch to effective duration?
How should I think about the relationship between Macaulay duration and modified duration instead of memorizing two separate definitions?
Why do hedge calculations often use dollar duration or DV01 instead of just modified duration?
When should I prefer historical simulation VaR over delta-normal VaR?
Join the Discussion
Ask questions and get expert answers.