Diversification Benefit and Its Attribution

What Diversification Benefit Measures

When risks are combined, aggregate capital may be less than the sum of the capital required for each risk on its own. The difference is the portfolio’s diversification benefit. It quantifies a reduction in capital under a specified risk measure and model; it is not, by itself, a description of which business unit caused that reduction.

The comparison is meaningful only when standalone and aggregate capital use a consistent basis: the same horizon, confidence level, valuation basis, and risk measure. If these differ, the subtraction can mix unlike quantities rather than measure diversification.

Definition: For portfolio units $1,\ldots,n$, let $X_i$ be the loss associated with unit $i$, and let $\rho$ be the chosen capital risk measure. The standalone capital of unit $i$ is $c_i=\rho(X_i)$. The aggregate capital is $C=\rho(\sum_{i=1}^n X_i)$. The diversification benefit is $$ D=\sum_{i=1}^n c_i-C. $$ When $\sum_i c_i>0$, the diversification-benefit proportion, using total standalone capital as the denominator, is $$ d=\frac{D}{\sum_{i=1}^n c_i}. $$

This tutorial uses total standalone capital as the denominator when it reports a proportion. Other ratios can be useful for other purposes, but they must be named explicitly: for example, $D/C$ is a ratio to aggregate capital, not the proportion of standalone capital saved. Also, $D$ need not be positive for every risk measure or model. A negative value means aggregate capital exceeds the sum of the standalone amounts on the specified basis.

Calculate the Benefit Before Attributing It

A portfolio has one total diversification benefit, but there is generally no unique way to assign that total to its units. An attribution requires a stated allocation method. If $K_i$ is the capital allocated to unit $i$, its attributed benefit under that allocation is $b_i=c_i-K_i$. When the allocations sum to aggregate capital, the attributed benefits sum to the portfolio benefit. This identity does not select an allocation method or imply that the attribution is unique.

\sum_{i=1}^n b_i =\sum_{i=1}^n(c_i-K_i) =\sum_{i=1}^n c_i-\sum_{i=1}^n K_i =\sum_{i=1}^n c_i-C =D \quad\text{if}\quad \sum_{i=1}^n K_i=C.

The distinction matters in practice. The portfolio benefit is determined by the standalone and aggregate capital figures. An attributed benefit additionally depends on how the aggregate capital is allocated. Two allocation methods can reconcile to the same aggregate capital and total diversification benefit while assigning different amounts to individual units.

Worked Example: Measuring Benefit From Reported Capital

A hypothetical insurer measures two business units using the same annual-horizon capital measure. Unit A has standalone capital of $95 million, Unit B has standalone capital of $65 million, and the combined portfolio has aggregate capital of $132.50 million. Calculate the portfolio diversification benefit and its proportion of standalone capital.

Method. Add the standalone capital first, then subtract aggregate capital. Use total standalone capital as the proportion’s denominator.

The total standalone capital is $$ 95+65=160\text{ million dollars}. $$ The diversification benefit is $$ D=160-132.50=27.50\text{ million dollars}. $$ The proportion is $$ d=\frac{27.50}{160}=0.171875. $$ Rounded to two decimal places as a percentage, the benefit is 17.19% of total standalone capital. The portfolio therefore has $27.50 million less capital than the sum of the units’ standalone amounts under this measure.

Check. Aggregate capital plus benefit returns the standalone total: $132.50+27.50=160$ million. The percentage is below 100%, consistent with a benefit smaller than the standalone total.

Common mistake avoided. Using $27.50/132.50=20.75\%$ would calculate benefit relative to aggregate capital. That ratio is mathematically valid, but it is not the standalone-capital proportion requested here.

Covariance Aggregation: A Derivation and Application

A commonly used aggregation model is based on variances and pairwise correlations. It applies when the portfolio capital measure is proportional to the standard deviation of the modeled losses, and the covariance structure is suitable for the intended aggregation. In particular, it is not a general formula for aggregating arbitrary risk measures or for losses with dependence that the covariance model does not represent.

For two losses $X_1$ and $X_2$, the variance of their sum follows from expanding the centered square:

\begin{aligned} \operatorname{Var}(X_1+X_2) &=E\left[\left((X_1-E[X_1])+(X_2-E[X_2])\right)^2\right]\\ &=\operatorname{Var}(X_1)+\operatorname{Var}(X_2) +2\operatorname{Cov}(X_1,X_2). \end{aligned}

If $\sigma_i$ is the standard deviation of $X_i$, and $\rho_{12}$ is their correlation, then $\operatorname{Cov}(X_1,X_2)=\rho_{12}\sigma_1\sigma_2$. Thus the portfolio standard deviation is the square root of the summed variances and twice the covariance. If capital is $k$ times standard deviation for the same positive constant $k$ at both the standalone and portfolio levels, the same aggregation applies to standalone capital amounts $c_i=k\sigma_i$:

C=\sqrt{c_1^2+c_2^2+2\rho_{12}c_1c_2}.

The formula requires a valid correlation, $-1\leq\rho_{12}\leq1$, and consistent measurement of both units. For multiple units, the corresponding expression includes every pairwise covariance. A correlation value alone does not determine the result unless the standalone amounts and the aggregation assumptions are also specified.

Worked Example: Covariance Aggregation and Relative Benefit

A hypothetical portfolio has two units with standalone capital amounts of $60 million and $80 million. Assume capital is proportional to standard deviation under the same model for each unit and the portfolio. The modeled correlation between losses is $0.25$. Calculate aggregate capital, the diversification benefit, and its proportion of standalone capital. Report capital to the nearest cent in millions and the percentage to two decimal places.

Method. Apply the covariance aggregation formula, then use the standalone total as the denominator.

First calculate the squared standalone amounts and covariance term: $$ 60^2=3{,}600,\qquad 80^2=6{,}400,\qquad 2(0.25)(60)(80)=2{,}400. $$ Therefore, $$ C=\sqrt{3{,}600+6{,}400+2{,}400} =\sqrt{12{,}400} =111.355287\ldots\text{ million dollars}. $$ Rounded to the nearest cent in millions, aggregate capital is $111.36 million. The standalone total is $$ 60+80=140\text{ million dollars}. $$ Using the unrounded aggregate result before rounding the final answer, $$ D=140-\sqrt{12{,}400}=28.644712\ldots\text{ million dollars}, $$ or $28.64 million to the nearest cent in millions. The proportion is $$ d=\frac{28.644712\ldots}{140}=0.2046056\ldots, $$ which is 20.46% to two decimal places.

Check. $111.355287\ldots+28.644712\ldots=140$. The result also lies between the larger standalone amount, $80 million, and the sum, $140 million, as expected for this positive-correlation example.

Common mistake avoided. Simply adding the standalone amounts gives $140 million, which is the no-diversification sum, not the covariance-aggregated capital. The correlation term must be included before taking the square root.

Attributing the Benefit Under a Stated Allocation

Once an allocation has been selected, attribution is a subtraction for each unit: allocated capital is compared with that unit’s standalone capital. A unit allocated less than its standalone capital is assigned a positive benefit under that method. The allocation method determines the unit-level amounts; the total benefit remains the same as long as the allocated amounts reconcile to aggregate capital.

For example, one possible method is to allocate aggregate capital in proportion to standalone capital. Under this method, unit $i$ receives $K_i=Cc_i/\sum_j c_j$. The attributed benefits then follow from $b_i=c_i-K_i$. This is an allocation convention, not a conclusion that each unit independently generated a particular share of diversification.

Worked Example: Attributing a Known Portfolio Benefit

A hypothetical three-unit portfolio has standalone capital of $120 million, $90 million, and $60 million. Aggregate capital is $216 million. A specified allocation method assigns $100 million, $72 million, and $44 million to the units, respectively. Calculate the portfolio benefit, each unit’s attributed benefit, and each unit’s share of the total benefit. Report shares to one decimal place.

Method. First verify that allocated capital reconciles to aggregate capital. Then subtract each allocation from its corresponding standalone amount.

The standalone total is $$ 120+90+60=270\text{ million dollars}. $$ Thus the portfolio benefit is $$ D=270-216=54\text{ million dollars}. $$ The allocations reconcile: $$ 100+72+44=216\text{ million dollars}. $$ The unit-level attributed benefits are $$ b_1=120-100=20,\qquad b_2=90-72=18,\qquad b_3=60-44=16 $$ in millions of dollars. Their sum is $20+18+16=54$ million dollars, matching the portfolio benefit.

The shares of the total benefit are $$ \frac{20}{54}=37.037\ldots\%,\qquad \frac{18}{54}=33.333\ldots\%,\qquad \frac{16}{54}=29.629\ldots\%. $$ Rounded to one decimal place, the shares are 37.0%, 33.3%, and 29.6%. They sum to 99.9% after rounding; before rounding, they sum to 100%.

Check. The allocated amounts total $216 million, and the attributed benefits total $54 million. Together, these reproduce the $270 million standalone total.

Common mistake avoided. Treating the $54 million portfolio benefit as though it were automatically assigned in proportion to standalone capital would produce different unit-level values. The problem specifies an allocation; the attributed benefits must be calculated from that allocation.

Interpreting Comparisons and Avoiding Traps

Diversification-benefit amounts and proportions answer different questions. The amount $D$ is measured in capital units and reports the absolute reduction from the standalone sum. The proportion $d$ scales that reduction by standalone capital, making portfolios of different sizes easier to compare, provided they use consistent definitions and assumptions.

A larger absolute benefit does not necessarily mean a larger relative benefit. A larger portfolio can have a greater dollar reduction but a smaller percentage reduction. Similarly, comparing two reported proportions is meaningful only when the denominators and underlying capital bases match.

Key takeaway: Calculate the portfolio benefit from the standalone total and aggregate capital first. Report its denominator whenever giving a proportion. Attribute the total to units only after identifying the allocation method, and check that allocated capital reconciles to aggregate capital.

An exam-style calculation often turns on one of three distinctions:

Standalone sum versus aggregate capital. The standalone sum is the benchmark; it is not the diversified portfolio result.
Benefit versus benefit proportion. The amount is a difference in capital units. The proportion requires a stated denominator.
Portfolio result versus unit attribution. The portfolio benefit is determined by the capital totals; the unit-level attribution depends on the chosen allocation.

These checks keep a sound calculation from becoming a misleading interpretation. They also align with the CAS Exam 9 (Risk Management for Actuaries) syllabus distinction between measuring portfolio capital and allocating it.

Check Your Understanding

For each problem, use total standalone capital as the denominator when asked for a diversification-benefit proportion. Capital amounts are hypothetical.

A portfolio has standalone capital of $48 million and $32 million, with aggregate capital of $68 million. Find the benefit and its proportion of standalone capital. (Answer: $12 million; 15.00%)
Two units have standalone standard-deviation-based capital of $30 million and $40 million, with modeled loss correlation $0.50$. Assume the same proportionality between standard deviation and capital at standalone and portfolio levels. Find aggregate capital, benefit, and benefit proportion. Round capital to the nearest cent in millions and the percentage to two decimal places. (Answer: Aggregate capital $60.83 million; benefit $9.17 million; proportion 13.10%)
Three units have standalone capital of $50 million, $40 million, and $30 million; their portfolio aggregate capital is $102 million. Calculate the benefit and its proportion of the standalone total. (Answer: $18 million; 15.00%)
Three units have standalone capital of $70 million, $60 million, and $70 million. A specified allocation assigns them $62 million, $51 million, and $63 million. Find the total benefit and each unit’s attributed benefit. (Answer: Total benefit $24 million; attributed benefits $8 million, $9 million, and $7 million)
Portfolio A has standalone total capital of $300 million and aggregate capital of $225 million. Portfolio B has standalone total capital of $260 million and aggregate capital of $210 million. Which has the larger absolute benefit, and which has the larger benefit proportion? (Answer: A has the larger absolute benefit, $75 million versus $50 million; A also has the larger proportion, 25.00% versus 19.23%)

What You'll Learn