Diagnosing Premium Mix Effects

Why Premium Mix Matters

An insurer can report a higher average premium even when it has made no rate change. For example, its portfolio may contain a larger share of policies in segments that have higher premiums per policy or per exposure. Conversely, a shift toward lower-premium segments can offset some or all of an increase in segment rates. To diagnose an aggregate premium change, separate changes in the premium charged within segments from changes in the distribution of business across those segments.

The key is to compare like with like. Choose a unit of exposure or policy count, divide business into consistent segments, and calculate each segment’s average premium on that same basis. Then the aggregate average premium is a weighted average. The weights describe the portfolio mix; they are not the segment premium amounts.

Definition: Let $j=1,\ldots,J$ index mutually exclusive and exhaustive segments. Let $w_j$ be segment $j$'s share of total exposure, so $w_j\geq 0$ and $\sum_j w_j=1$. Let $p_j$ be average premium per unit of that exposure in segment $j$. The aggregate average premium per exposure is $P=\sum_j w_jp_j$. A mix effect is the part of a change in $P$ associated with changes in the weights $w_j$, rather than changes in the segment premiums $p_j$.

The exposure basis must be specified. Weights might be shares of policies, earned car-years, or another appropriate exposure measure. A policy-count-weighted average is not interchangeable with a premium-weighted average or an earned-exposure-weighted average. Use the same definition of segment, exposure, and premium in both periods; otherwise, the comparison may combine mix changes with changes in measurement.

Separating Rate and Mix Changes

Use subscripts 0 and 1 for the earlier and later periods. Thus, $w_{0j}$ and $p_{0j}$ are the earlier weight and segment premium, and $w_{1j}$ and $p_{1j}$ are their later values. The two aggregate averages are $P_0=\sum_j w_{0j}p_{0j}$ and $P_1=\sum_j w_{1j}p_{1j}$.

One common convention is the rate-first decomposition: calculate the rate effect using the earlier mix, then calculate the mix effect using the later segment premiums. The exact identity is:

P_1-P_0 = \underbrace{\sum_j w_{0j}(p_{1j}-p_{0j})}_{\text{rate effect at earlier mix}} + \underbrace{\sum_j (w_{1j}-w_{0j})p_{1j}}_{\text{mix effect at later segment premiums}}.

This decomposition assigns the interaction between changing rates and changing mix to the mix effect. Another valid convention calculates the mix effect at earlier segment premiums and the rate effect at later weights:

P_1-P_0 = \sum_j (w_{1j}-w_{0j})p_{0j} + \sum_j w_{1j}(p_{1j}-p_{0j}).

Both decompositions reconcile exactly to the same total change. Their individual rate and mix components can differ, so state which convention is being used. Neither convention is inherently wrong; the important requirements are consistency and clear interpretation.

Fix the comparison basis.
Choose segments and an exposure unit that are defined consistently in both periods.

Calculate weights and segment premiums.
For each period, confirm that the segment weights sum to 1 and calculate the average premium per exposure within each segment.

Calculate the aggregate averages.
Apply $P=\sum_j w_jp_j$ separately to each period. Do not average segment premiums without applying the weights.

Decompose and reconcile.
Apply one stated convention for rate and mix effects, then verify that their sum equals $P_1-P_0$.

Why the Decomposition Is Exact

The rate-first identity follows by adding and subtracting an intermediate portfolio average: the later segment premiums evaluated at the earlier weights, $\sum_j w_{0j}p_{1j}$. This is a comparison, not necessarily a separately observed portfolio.

\begin{aligned} P_1-P_0 &=\sum_j w_{1j}p_{1j}-\sum_j w_{0j}p_{0j}\\ &=\left(\sum_j w_{0j}p_{1j}-\sum_j w_{0j}p_{0j}\right) +\left(\sum_j w_{1j}p_{1j}-\sum_j w_{0j}p_{1j}\right)\\ &=\sum_j w_{0j}(p_{1j}-p_{0j}) +\sum_j(w_{1j}-w_{0j})p_{1j}. \end{aligned}

The first grouped difference holds the mix at its earlier value and changes the segment premiums. The second holds segment premiums at their later values and changes the mix. Since the intermediate terms cancel, the components always add to the observed aggregate difference. The identity requires weights that sum to 1 in each period and segment premiums measured on a consistent basis; it does not require a particular distribution of risks within a segment.

Worked Example: A Rate Increase and a Shift Toward Higher-Premium Business

A hypothetical insurer has two segments. Initially, 50% of exposures are in Segment A, with average premium of $200 per exposure, and 50% are in Segment B, with average premium of $600. Later, both segment premiums rise by 5%, while the exposure shares change to 25% in A and 75% in B. Calculate the initial and later aggregate averages, the rate and mix effects using the rate-first convention, and the total percentage change.

Method and initial average. The weights sum to 1 in each period, and premiums are per exposure. The initial average is:

P_0=0.50(200)+0.50(600)=100+300=\$400.

Update the segment premiums. A 5% increase gives $p_{1A}=200(1.05)=\$210$ and $p_{1B}=600(1.05)=\$630$. The later average, using the later weights, is:

P_1=0.25(210)+0.75(630)=52.50+472.50=\$525.

Rate effect at earlier mix. The segment increases are $210-200=\$10$ and $630-600=\$30$. Apply the earlier 50%-50% weights:

0.50(10)+0.50(30)=5+15=\$20.

Mix effect at later segment premiums. The weight changes are $0.25-0.50=-0.25$ for A and $0.75-0.50=0.25$ for B:

(-0.25)(210)+(0.25)(630)=-52.50+157.50=\$105.

Reconcile and calculate the percentage change. The effects sum to $20+105=\$125$, which equals $525-400=\$125$. Relative to the initial average, the increase is $125/400=0.3125$, or 31.25%. The rate change within each segment is 5%, but the aggregate average increases more because the later portfolio has a larger share in the higher-premium segment. The common mistake this method avoids is reporting the later average as $630: that is Segment B’s premium, not the weighted average across the portfolio.

Mix Changes With No Segment Rate Change

If segment premiums are unchanged, the aggregate change is entirely due to mix. That does not mean any individual policyholder’s premium changed. It means the portfolio average changed because the relative amounts of business in the segments changed. This distinction is useful when interpreting trends in written or earned premium: an aggregate increase alone does not establish that rates increased.

Worked Example: Average Premium Rises While Segment Premiums Stay Fixed

A hypothetical equipment insurer records average annual premiums per policy of $300 in Segment A and $900 in Segment B. Initially, 80% of its policies are in A and 20% in B. In a later period the shares are 50% and 50%, while the premiums in both segments are unchanged. Find the average premium per policy in each period, the dollar and percentage changes, and the explanation.

Initial average. Weight each segment premium by its policy share:

P_0=0.80(300)+0.20(900)=240+180=\$420.

Later average. The new mix gives:

P_1=0.50(300)+0.50(900)=150+450=\$600.

Change and check. The average rises by $600-420=\$180$, and the percentage increase is $180/420=0.428571\ldots$, or 42.86% rounded to two decimals. Because the segment premiums did not change, this $180 increase is a mix effect, not a rate increase. A check using the mix-only formula gives $(0.50-0.80)(300)+(0.50-0.20)(900)=-90+270=\$180$. The method avoids attributing an aggregate premium change to rate action when the segment rates are constant.

Mix Effects, Total Premium, and Decomposition Choice

An average premium per exposure is not the same as total premium. If total exposure changes, total premium can change even when the average and the mix do not. With total exposures $E$, total premium is $T=EP$. Therefore, a change in $T$ can reflect exposure volume, average premium, or both. When volume and mix both change, calculate them separately rather than interpreting total premium growth as a rate change.

The rate-first and mix-first decompositions also differ in how they assign the interaction between rate and mix changes. The following example calculates both and verifies that each reconciles to the same total.

Worked Example: Comparing Rate-First and Mix-First Results

A hypothetical insurer has two segments. Initially, the weights are 80% and 20%, and the average premiums are $300 and $900. Later, the weights are 50% and 50%, and the segment premiums have each increased by 10%, to $330 and $990. Calculate the aggregate change under both decomposition conventions and explain why the components differ.

Calculate the aggregate averages. Initially, $P_0=0.80(300)+0.20(900)=240+180=\$420$. Later, $P_1=0.50(330)+0.50(990)=165+495=\$660$. Thus, the aggregate average rises by $660-420=\$240$, or $240/420=57.14\%$ rounded to two decimals.

Rate-first decomposition. At the earlier mix, the premium increases are $30 and $90, so the rate effect is $0.80(30)+0.20(90)=24+18=\$42$. The weight changes are $-0.30$ and $0.30$; using later premiums, the mix effect is $(-0.30)(330)+(0.30)(990)=-99+297=\$198$. Their sum is $42+198=\$240$.

Mix-first decomposition. Using the earlier premiums, the mix effect is $(-0.30)(300)+(0.30)(900)=-90+270=\$180$. At the later mix, the rate effect is $0.50(30)+0.50(90)=15+45=\$60$. Their sum is $180+60=\$240$.

Both calculations reconcile, but the rate-first mix effect is $18 larger, and its rate effect is $18 smaller. The difference is the interaction assigned differently by the two conventions. The common mistake is to compare components from different conventions as if they must be identical. The total aggregate change is the same; state the convention and check the sum.

Common Diagnostic Traps

Using equal weights without justification. Segment premiums must be weighted by the portfolio shares on the chosen exposure basis, not simply averaged.
Confusing mix with rate change. A higher aggregate average can result from more business in high-premium segments even when segment premiums are unchanged.
Changing the exposure basis. A policy share in one period cannot be compared directly with an earned-exposure share in another without addressing the basis difference.
Ignoring the total exposure count. A change in total premium dollars can arise from growth or contraction in exposure volume as well as from changes in average premium.
Failing to identify the decomposition convention. Rate and mix components depend on whether the calculation uses earlier or later values for the other factor; the overall change does not.

A mix analysis describes the arithmetic of an aggregate premium measure; by itself, it does not establish whether the selected rates are adequate or whether the changed portfolio has the same risk characteristics within each segment. Interpret the result in the context of the comparison basis and the purpose of the analysis. This distinction is aligned with the CAS Exam 5 (Basic Ratemaking and Estimated Claim Liabilities) syllabus.

Check Your Understanding

For each problem, use the stated exposure basis and show enough intermediate work to verify both the aggregate average and the change.

A hypothetical insurer’s premiums per exposure are $150 in Segment A and $450 in Segment B. The initial exposure shares are 60% and 40%; later they are 30% and 70%. Segment premiums do not change. Find both aggregate averages and the mix effect. (Answer: initial $270; later $360; mix effect +$90 per exposure, or +33.33%.)
Initially, 75% of exposures are in Segment A and 25% in B, with premiums of $240 and $640. Both premiums increase by 7.5%, and later weights are 50% and 50%. Using the rate-first convention, calculate the aggregate change, rate effect, mix effect, and percentage change. (Answer: later premiums $258 and $688; initial average $340; later average $473; rate effect $25.50; mix effect $107.50; total +$133, or +39.12%.)
A hypothetical portfolio has 80 policies in Segment A and 20 in B, with unchanged premiums of $400 and $1,000. Later it has 60 policies in each segment. Calculate total premium in both periods, average premium per policy in both periods, and identify the mix and volume changes. (Answer: initial total $52,000 and average $520; later total $84,000 and average $700; total premium increases $32,000, or 61.54%, because policy count increases 20% and the average rises 34.62% from mix.)
Initially, segment weights are 40% and 60%, and segment premiums are $200 and $500. Later, weights are 70% and 30%, and premiums are $220 and $550. Under the rate-first convention, calculate the initial and later average, both effects, and verify the total change. (Answer: initial $380; later $319; rate effect +$38; mix effect −$99; total change −$61 per exposure.)

What You'll Learn