5.4 Risk Parity & Equal Risk Contribution

Basic Prerequisites: Linear Algebra & Matrices and Calculus & Convex Optimization. (these are the folder-level prerequisites for pages 02–06; page 01 states its own, smaller, entry requirements)

1. Intuition & Practical Objective

Markowitz mean-variance optimization balances dollars against an expected-return forecast; this folder’s discipline - risk budgeting, risk parity and equal-risk-contribution (ERC) - balances risk. The founding observation (Qian 2005) is that diversification measured in dollars is not diversification measured in risk: because equities are ~3–4× as volatile as government bonds, a “balanced” 60/40 stock/bond portfolio is secretly a ~90/10 equity-risk portfolio (in the worked numbers below, equity carries 92.7% of total risk). Diversification on paper, concentration in practice.

The fix is to equalize the risk each asset contributes to the portfolio, not the capital each receives. The two names in the title are the same idea at two levels of generality:

Risk parity (naive): make each asset’s marginal contribution to risk equal by weighting inversely to volatility - $w_{i} \propto 1/ σ_{i}$ . Correct only when all correlations are equal.
Equal risk contribution (ERC): make each asset’s total risk contribution equal, $R C_{i} = σ (w) / n$ , which correctly folds in the covariance structure $Σ$ . This is the disciplined form of risk parity, and ERC = risk parity under a constant-correlation $Σ$ .

This is the hub: it gives you the fast formula lookup (job #1) and routes you to six sub-pages that walk from raw intuition through the risk decomposition, the ERC problem and its closed forms, general risk budgeting, failure modes, and the leverage-aversion debate. Every formula below is transcribed from Maillard, Roncalli & Teïletche (2010) and Qian (2005, 2006) and cross-checked against Asness, Frazzini & Pedersen (2012); the check-column numbers were re-executed and reproduced exactly from the verified corpus (§3).

The one-sentence essence. “Split the portfolio’s total risk - not its capital - into equal (or user-chosen) slices; because risk is a homogeneous function of the weights, Euler’s theorem forces $σ (w) = \sum_{i} w_{i} (Σ w)_{i} / σ (w)$ , so every unit of portfolio volatility always decomposes into $N$ well-defined per-asset risk budgets that add up to 100%.“

2. Mathematical Ground Truth & Derivations

Quick-Reference Lookup (job #1). Worked on the folder’s worked universe - the four assets of Maillard et al. (2010): volatilities $10%, 20%, 30%, 40%$ with the correlation matrix

ρ = 1.00 0.80 00 0.80 1.00 00 00 1.00 - 0.50 00 - 0.50 1.00 .

Notation: $w$ weights ( $\sum w_{i} = 1$ ), $Σ$ covariance matrix, $σ (w) = w^{⊤} Σ w$ portfolio volatility, $(Σ w)_{i}$ the $i$ -th row of the matrix–vector product.

Quantity	Formula	Verified check
Portfolio volatility	$σ (w) = w^{⊤} Σ w$	ERC portfolio $σ = 0.1029$
Marginal risk contribution (MRC)	$MRC_{i} = \frac{\partial σ}{\partial w _{i}} = \frac{( Σ w ) _{i}}{w ^{⊤} Σ w}$	ERC: $[0.0671, 0.1342, 0.1061, 0.1414]$
Total risk contribution (RC)	$R C_{i} = w_{i} \frac{( Σ w ) _{i}}{w ^{⊤} Σ w}$	ERC: $0.0257$ each, all equal
Percentage risk contribution	$R C_{i} / σ (w)$	ERC: 25.0% for each of the 4 assets
Euler decomposition	$σ (w) = \sum_{i = 1}^{N} R C_{i}$	$\sum R C_{i} = 0.102934 = σ$ exactly (verified < $1 0^{- 9}$ )
ERC condition	$R C_{i} = R C_{j} = σ (w) / N ⟺ w_{i} (Σ w)_{i} = w_{j} (Σ w)_{j}$	ERC weights $[0.384, 0.192, 0.243, 0.182]$
Two-asset ERC (closed form)	$w_{1} = \frac{σ _{1}^{- 1}}{σ _{1}^{- 1} + σ _{2}^{- 1}} = \frac{σ _{2}}{σ _{1} + σ _{2}}, w_{2} = \frac{σ _{2}^{- 1}}{σ _{1}^{- 1} + σ _{2}^{- 1}} = \frac{σ _{1}}{σ _{1} + σ _{2}}$ (independent of $ρ$ )	$σ_{1} = 15.1%, σ_{2} = 4.6%, ρ = 0.2 \Rightarrow w = [0.234, 0.766]$
Constant-correlation ERC	$w_{i} = \frac{σ _{i}^{- 1}}{\sum _{j} σ _{j}^{- 1}}$ (= inverse-vol)	vols $10/20/30/40$ , $ρ = 0.3$ : $[0.48, 0.24, 0.16, 0.12]$
Beta form	$w_{i} \propto β_{i}^{- 1}, β_{i} = \frac{( Σ w ) _{i}}{σ ( w ) ^{2}}$ (endogenous)	useful for interpretation, not closed form
Risk budgeting (budgets $b$ )	$R C_{i} (w) = b_{i} σ (w), \sum b_{i} = 1$ (ERC = $b_{i} = 1/ N$ )	$b = [.40, .30, .20, .10] \Rightarrow w = [0.493, 0.190, 0.194, 0.123]$

The one formula to never forget. $R C_{i} = w_{i} (Σ w)_{i} / σ (w)$ . It is the product of how much you hold ( $w_{i}$ ) and how risky that holding is to the whole portfolio ( $(Σ w)_{i} / σ (w)$ ), and these $N$ products always sum to the portfolio volatility. The naive inverse-vol weight $w_{i} \propto σ_{i}^{- 1}$ is not this formula’s equalizer - it ignores correlations.

3. Computational Implementation - the risk-budget engine

This runs on the standard library only (a cyclical coordinate-descent solver for the risk-budgeting program). It reproduces every verified number above: the ERC weights, the equal-RC verification, the exact Euler decomposition, the inverse-vol comparison, arbitrary risk budgets, and the two closed forms.

Read the row labelled inverse-vol and the row labelled ERC: same two-decimal $σ$ , yet inverse-vol concentrates 78% of the risk in two of the four assets (39.1%/39.1%/10.9%/10.9%) because it ignores the correlation structure - asset 1&2 are highly correlated ( $ρ = 0.8$ ) while asset 3 has a negative correlation ( $ρ = - 0.5$ ) with asset 4 that erc exploits as free diversification. ERC’s 25/25/25/25 is the true risk-equalizer.

4. Failure Modes & First-Principles Breakdowns

Hub signposts - the folder’s failure-mode analysis lives in 05 · Failure Modes & Practice. In one line each:

Garbage-in, garbage-out risk allocation. ERC is a pure function of $Σ$ ; a poorly-estimated covariance (short window, $N$ close to $T$ , no shrinkage/denoising) quietly hands the risk budget to the wrong assets - the connector to Covariance Shrinkage & RMT Denoising.
Correlation-regime flips. ERC is built on a point estimate of $ρ$ ; a stock–bond correlation that was $- 0.2$ and becomes $+ 0.5$ (2022) invalidates every budget computed from the old $Σ$ .
The leverage trap. A prudent risk-balanced portfolio has far more bond risk-budget than dollar-budget; targeting 60/40 returns forces 2–3× leverage on the bond leg, which is exactly what blows up when rates rise and both legs fall together.

5. References

Maillard, Sébastien; Roncalli, Thierry & Teïletche, Jérôme: The Properties of Equally Weighted Risk Contribution Portfolios, Journal of Portfolio Management 36(4):60–70 (2010)
Qian, Edward: Risk Parity Portfolios: Efficient Portfolios Through True Diversification, PanAgora Asset Management (2005)
Qian, Edward: On the Financial Interpretation of Risk Contribution: Risk Budgets Do Add Up, Journal of Investment Management 4(4) (2006)
Asness, Clifford; Frazzini, Andrea & Pedersen, Lasse H. (with Black 1972): Leverage Aversion and Risk Parity, Financial Analysts Journal 68(1):47–59 (2012)
Roncalli, Thierry: Introduction to Risk Parity and Budgeting, Chapman & Hall/CRC (2013)

6. Connected Graph Bridges

Foundational base: Linear Algebra · Calculus & Convex Optimization · Statistics
Sub-pages (in-folder): 01 From Zero · 02 Risk Contributions · 03 Equal Risk Contribution · 04 Risk Budgeting · 05 Failure Modes · 06 Advanced Extensions
Sibling topics: Mean–Variance & Markowitz · Covariance Shrinkage & RMT Denoising · Hierarchical Risk Parity (HRP) · Transaction Costs

Beginner: start at 01 · Practitioner: start at 05

Kwant Atlas

Explorer

5.4 Risk Parity & Equal Risk Contribution

1. Intuition & Practical Objective

2. Mathematical Ground Truth & Derivations

3. Computational Implementation - the risk-budget engine

4. Failure Modes & First-Principles Breakdowns

5. References

6. Connected Graph Bridges

5.4.1 Risk Parity & ERC from Zero

5.4.2 Risk Contributions

5.4.3 Equal Risk Contribution (ERC)

5.4.4 Risk Budgeting

5.4.5 Failure Modes & Practice

5.4.6 Advanced Extensions