Incorporating Stabilizing Ballast

Why a Stabilizing Ballast Is Used

An experience modification compares observed losses with expected losses. When the observed experience differs substantially from expectation, an unadjusted ratio can produce a large change in the factor. A stabilizing ballast moderates that change: it gives weight to the observed experience while also retaining weight on a neutral factor of 1.00. The result is less responsive to the observed loss deviation than the unballasted ratio.

This tutorial uses a supplied ballast as part of the experience modification calculation. The amount of ballast is treated as a plan input; it is not estimated from the losses in the examples unless the problem specifically asks us to solve for it. This material is aligned with the CAS Exam 8 (Advanced Ratemaking) syllabus.

Definition: Let $A$ be actual losses, $E$ be expected losses, and $B$ be a nonnegative stabilizing ballast, all measured on a consistent dollar basis. For $E>0$, the ballast modification is $$ M=\frac{A+B}{E+B}=1+\frac{A-E}{E+B}. $$ The ballast is added to both actual and expected losses in the ratio form. Equivalently, it reduces the weight assigned to the actual-to-expected experience ratio.

The neutral point is $M=1.00$. If actual losses equal expected losses, the factor is 1.00 for every permissible ballast amount. If actual losses exceed expected losses, the factor is above 1.00; if actual losses are below expected losses, it is below 1.00. The ballast reduces the distance from 1.00 in either direction.

Deriving the Credibility Interpretation

The formula can be understood as a weighted average of the observed actual-to-expected ratio and the neutral factor. Define the ballast-based weight $Z=E/(E+B)$. When $E>0$ and $B\geq 0$, $0<Z\leq 1$. The weight on the neutral factor is $1-Z=B/(E+B)$.

Start with the weighted average.
Weight the experience ratio $A/E$ by $Z$, and the neutral factor 1 by $1-Z$: $M=Z(A/E)+(1-Z)(1)$.

Substitute the ballast-based weight.
$M=[E/(E+B)](A/E)+[B/(E+B)]$.

Simplify.
The first term is $A/(E+B)$, so $M=A/(E+B)+B/(E+B)=(A+B)/(E+B)$.

Express the deviation from neutral.
Subtract 1: $M-1=[(A+B)-(E+B)]/(E+B)=(A-E)/(E+B)$. Thus $M=1+(A-E)/(E+B)$.

This derivation explains why the ballast is added to both terms in the ratio. It is equivalent to blending the observed ratio $A/E$ with 1.00. The actual-to-expected ratio is not itself multiplied by an arbitrary ballast; its weight is $Z$, while the neutral factor receives the remaining weight.

Calculating a Modification With Ballast

A reliable calculation starts by checking that the loss amounts use the same basis and dollar units. Then identify $A$, $E$, and $B$, apply the specified formula, and interpret the result relative to 1.00. Do not round intermediate amounts; round the final factor as requested.

Worked Example: Applying a Supplied Ballast

A hypothetical contractor has actual losses of $104,000, expected losses of $80,000, and a plan-specified ballast of $20,000. Calculate the modification and the ballast-based weight on the observed experience ratio. Round the final modification to four decimal places.

Method. Use $M=(A+B)/(E+B)$. The common mistake to avoid is adding ballast to the denominator but not to the numerator, which would not represent the stated ballast formula.

Step 1: Substitute the amounts. The numerator is $A+B=\$104{,}000+\$20{,}000=\$124{,}000$. The denominator is $E+B=\$80{,}000+\$20{,}000=\$100{,}000$.

Step 2: Calculate the modification.

M=\frac{\$124{,}000}{\$100{,}000}=1.24.

The final modification is 1.2400. For the credibility interpretation, $Z=E/(E+B)=\$80{,}000/\$100{,}000=0.80$. The unballasted actual-to-expected ratio is $A/E=\$104{,}000/\$80{,}000=1.30$. Therefore $0.80(1.30)+0.20(1.00)=1.04+0.20=1.24$, which checks the result.

A ballast does not change the underlying actual or expected losses. It changes how strongly their difference affects the indicated factor. In this example, the experience ratio of 1.30 is pulled toward 1.00, giving 1.24.

How Changing Ballast Changes the Factor

For fixed actual and expected losses, the direction of the ballast effect follows from the deviation form: $M-1=(A-E)/(E+B)$. Increasing $B$ makes the denominator larger while leaving the numerator $A-E$ unchanged. Consequently, the magnitude of the deviation from 1.00 decreases. If $A>E$, the factor decreases toward 1.00; if $A<E$, it increases toward 1.00.

The same conclusion follows by differentiating with respect to ballast:

\frac{\partial M}{\partial B} =\frac{(E+B)-(A+B)}{(E+B)^2} =\frac{E-A}{(E+B)^2}.

The denominator is positive for $E>0$ and $B\geq 0$. Thus the derivative is negative when $A>E$, positive when $A<E$, and zero when $A=E$. As $B$ becomes very large, the ratio $(A+B)/(E+B)$ approaches 1.00. A finite positive ballast therefore stabilizes the factor; it does not necessarily make the factor exactly 1.00.

Worked Example: Comparing Two Ballast Amounts

A hypothetical repair business has actual losses of $117,000 and expected losses of $150,000. Calculate the modification using ballast amounts of $30,000 and $60,000. Report both factors to four decimal places and quantify the change when the ballast increases.

Method. Calculate $(A+B)/(E+B)$ separately for each ballast. The common mistake to avoid is assuming that a higher ballast always lowers the factor. Here the unballasted experience is favorable, so added ballast raises the factor toward 1.00.

With $B=\$30{,}000$. The numerator is $\$117{,}000+\$30{,}000=\$147{,}000$; the denominator is $\$150{,}000+\$30{,}000=\$180{,}000$. Thus $M=\$147{,}000/\$180{,}000=0.816666\ldots$, or 0.8167.

With $B=\$60{,}000$. The numerator is $\$117{,}000+\$60{,}000=\$177{,}000$; the denominator is $\$150{,}000+\$60{,}000=\$210{,}000$. Thus $M=\$177{,}000/\$210{,}000=0.842857\ldots$, or 0.8429.

Change and check. Using unrounded values, the increase is $0.842857\ldots-0.816666\ldots=0.026190\ldots$, or 0.0262 to four decimal places. Both factors are below 1.00, and the larger ballast produces the factor closer to 1.00. This direction is consistent with the derivative because $A-E=\$117{,}000-\$150{,}000=-\$33{,}000$.

Solving for a Required Ballast

Sometimes the factor is specified and the unknown is the ballast. Provided the target factor differs from 1.00, rearrange the formula algebraically. Starting from $M=(A+B)/(E+B)$, multiply both sides by $E+B$, collect the terms involving $B$, and solve:

M(E+B)=A+B \quad\Longrightarrow\quad B(M-1)=A-ME \quad\Longrightarrow\quad B=\frac{A-ME}{M-1}.

This rearrangement requires $M\ne 1$. If $M=1$, the original equation requires $A=E$; when $A=E$, every nonnegative ballast gives 1.00, and when $A\ne E$, no finite ballast gives exactly 1.00. After solving, check that the resulting ballast is nonnegative if the problem describes a valid stabilizing ballast, and substitute it back into the original formula.

Worked Example: Finding Ballast for a Target Modification

A hypothetical equipment installer has actual losses of $90,000 and expected losses of $60,000. Determine the nonnegative ballast that produces a modification of 1.20. Verify the result using both the ratio formula and the credibility interpretation.

Method. Solve $1.20=(\$90{,}000+B)/(\$60{,}000+B)$. The common mistake to avoid is solving for ballast from the deviation formula but losing the sign on $M-1$; substitution into the original equation is an essential check.

Step 1: Clear the denominator. $1.20(\$60{,}000+B)=\$90{,}000+B$, so $\$72{,}000+1.20B=\$90{,}000+B$.

Step 2: Isolate $B$. Subtract $B$ and subtract $72,000 from both sides: $0.20B=\$18{,}000$. Therefore $B=\$18{,}000/0.20=\$90{,}000$. The result is nonnegative.

Step 3: Verify by substitution. The numerator is $\$90{,}000+\$90{,}000=\$180{,}000$, and the denominator is $\$60{,}000+\$90{,}000=\$150{,}000$. Their ratio is $1.20$, as required.

Step 4: Verify the credibility interpretation. The observed ratio is $\$90{,}000/\$60{,}000=1.50$. The weight is $Z=\$60{,}000/(\$60{,}000+\$90{,}000)=0.40$. The weighted factor is $0.40(1.50)+0.60(1.00)=0.60+0.60=1.20$. Both checks agree.

Assumptions, Interpretation, and Common Traps

The formula is meaningful when actual losses, expected losses, and ballast are defined on a consistent basis. They must refer to the same experience period and loss measure, and be expressed in the same dollar units. The expected loss amount must be positive for the actual-to-expected ratio and the credibility interpretation to be defined. A zero or negative expected amount is not handled by this formula; use the applicable plan instructions rather than forcing a ratio calculation.

A supplied ballast should be used as specified. It is not an additional loss, premium, or exposure, and it is not a value to choose solely because it produces a preferred factor. If an exercise asks how different ballast amounts affect the result, hold $A$ and $E$ fixed and recalculate. If it asks for the ballast that attains a target, solve algebraically and verify that the result is admissible.

Adding ballast only to the denominator: This changes the formula and does not produce the stated credibility blend. Use $(A+B)/(E+B)$.
Assuming the factor always falls as ballast rises: It falls toward 1.00 when $A>E$, but rises toward 1.00 when $A<E$.
Mixing units: If $A$ and $E$ are stated in thousands of dollars, convert $B$ to thousands before using it, or convert all three amounts to dollars.
Rounding too early: Retain full precision through comparisons and differences; round only the final reported factor to the requested precision.

Key takeaway: A stabilizing ballast moderates the deviation from 1.00 according to $M-1=(A-E)/(E+B)$. Larger ballast reduces the magnitude of that deviation, while the direction of movement depends on whether actual losses are above or below expected losses.

Check Your Understanding

For each problem, use the stated ballast formula and check whether the result moves in the expected direction relative to 1.00.

A hypothetical landscaping firm has actual losses of $50,000, expected losses of $40,000, and ballast of $10,000. Calculate its modification. (Answer: 1.2000)
A hypothetical courier has actual losses of $90,000, expected losses of $120,000, and ballast of $30,000. Calculate its modification. (Answer: 0.8000)
A hypothetical manufacturer has actual losses of $132,000 and expected losses of $110,000. Calculate the modifications at ballast amounts of $0, $22,000, and $44,000. (Answer: 1.2000, 1.1667, and 1.1429, respectively)
A hypothetical marina has actual losses of $105,000 and expected losses of $75,000. What nonnegative ballast produces a modification of 1.15? Verify by substitution. (Answer: $125,000; $(\$105{,}000+\$125{,}000)/(\$75{,}000+\$125{,}000)=1.15$)
A hypothetical commercial greenhouse has actual losses of $170,000 and expected losses of $200,000. Calculate the modification at ballast amounts of $50,000 and $100,000, then state the change. (Answer: 0.8800 and 0.9000, respectively; the factor increases by 0.0200 toward 1.00)

What You'll Learn