Trend-Adjusted Deductible Payments Across Policy Years

Why Price Units Matter Across Policy Years

When losses from different policy years are compared, the dollar amounts may reflect different price levels. A loss of $10,000 in an earlier year and a loss of $10,000 several years later do not necessarily represent the same underlying severity in constant dollars. A deductible also has a price level: a fixed nominal deductible can become smaller relative to the losses it applies to, while an indexed deductible may rise with the same trend as claim severity.

The calculation must therefore keep the loss and deductible in matching units. One direct method is to trend the base-year loss distribution and apply the deductible stated for that policy year. An equivalent method is to deflate that year’s deductible to base-year dollars, calculate the payment using the base-year distribution, and then—if a nominal payment is wanted—re-inflate the result.

Definition: Let $X$ be a nonnegative loss with a base-year severity distribution. For policy year $j$, suppose the loss is $X_j=c_jX$, where $c_j>0$ is a deterministic cumulative trend factor. If the ordinary deductible in policy-year dollars is $d_j$, the insurer payment per loss is $Y_j^L=(X_j-d_j)^+$, where $z^+=\max(z,0)$. Its expected nominal payment is $\operatorname{E}[Y_j^L]=c_j\operatorname{E}[(X-d_j/c_j)^+]$.

The factor $d_j/c_j$ is the deductible expressed in base-year dollars. The right-hand side first calculates the expected excess payment using base-year severity and a base-year deductible, then converts that expected amount back to policy-year dollars. If results are to be compared in base-year dollars, omit the final multiplication by $c_j$.

This model assumes the same base-year severity distribution underlies each policy year and that trend is proportional and deterministic. It does not, by itself, allow the underlying severity distribution to change for reasons other than the stated trend. It also does not discount payments for interest: discounting is a separate operation from adjusting for changes in claim costs.

The Payment Formula and Its Justification

For an ordinary deductible $d$, the insurer pays nothing when $X\le d$, and pays $X-d$ when $X>d$. Thus the payment is $(X-d)^+$. Under proportional trend, the payment in year $j$ is $(c_jX-d_j)^+$. Since $c_j$ is positive, it can be factored out:

(c_jX-d_j)^+ =c_j\left(X-\frac{d_j}{c_j}\right)^+.

Taking expectations gives the comparison formula. The deductible is divided by the trend factor because a policy-year dollar represents $1/c_j$ base-year dollars. The expected base-year-dollar payment is the nominal expected payment divided by $c_j$:

\operatorname{E}[(X_j-d_j)^+] =c_j\operatorname{E}\left[\left(X-\frac{d_j}{c_j}\right)^+\right], \qquad \text{base-year-dollar expectation} =\operatorname{E}\left[\left(X-\frac{d_j}{c_j}\right)^+\right].

For any nonnegative severity $X$, the expected payment above a deductible $a$ can also be written using the survival function $S_X(x)=P(X>x)$:

\operatorname{E}[(X-a)^+] =\int_a^\infty S_X(x)\,dx.

This identity follows by expressing the nonnegative payment as the area under its exceedance indicators: $(X-a)^+=\int_a^\infty \mathbf{1}_{\{X>x\}}\,dx$. Taking expectations gives the integral of $P(X>x)$. It is useful when the severity distribution is given by its survival function; the same price-unit rule still applies.

Key result: A fixed nominal deductible $d$ becomes $d/c_j$ in base-year dollars, so it becomes smaller in real terms as $c_j$ increases. If the deductible is indexed proportionally, $d_j=c_jd_0$, then $d_j/c_j=d_0$: the expected base-year-dollar payment is unchanged when the underlying base distribution is unchanged, while the nominal expected payment scales by $c_j$.

Worked Comparisons Across Policy Years

Worked Example: An Indexed Deductible and Exponential Severity

A hypothetical policy has base-year loss $X$ exponentially distributed with mean $10,000. In the next policy year, severity is trended by $c_1=1.10$, and the deductible is indexed from $2,000 to $d_1=$2,200$. Find the expected insurer payment in both base-year and policy-year dollars, and compare it with the base year.

Method. For an exponential severity with mean $\theta$, the survival function is $S_X(x)=e^{-x/\theta}$. Therefore $\operatorname{E}[(X-a)^+]=\int_a^\infty e^{-x/\theta}\,dx=\theta e^{-a/\theta}$. Here $\theta=$10,000$.

Base year. The base-year deductible is $2,000, so the expected payment is $$10,000e^{-2,000/10,000}=$10,000e^{-0.2}=$8,187.3075$. Rounded to cents, this is $8,187.31.

Next year in base-year dollars. The indexed deductible expressed in base-year dollars is $d_1/c_1=$2,200/1.10=$2,000$. The underlying severity and equivalent deductible are therefore unchanged, so the expected payment remains $8,187.31 in base-year dollars.

Next year in policy-year dollars. Multiply by the trend factor: $1.10($8,187.3075)=$9,006.0383$, or $9,006.04 to cents.

The base-year expected payment is unchanged, while the nominal expected payment rises by 10%. The calculation avoids treating an indexed deductible as though its value had stayed at $2,000 in policy-year dollars.

Worked Example: A Fixed Deductible Under Uniform Severity Trend

A hypothetical book has base-year severity $X$ uniform on $[0,$10,000]$. The nominal deductible remains $3,000 in both the base year and a later year, when severity is trended by $c=1.20$. Calculate expected insurer payments in each year, and compare the later payment in base-year dollars.

Method. If $X$ is uniform on $[0,b]$ and $0\le a\le b$, then $\operatorname{E}[(X-a)^+]=\frac{1}{b}\int_a^b(x-a)\,dx=\frac{(b-a)^2}{2b}$.

Base year. With $b=$10,000$ and $a=$3,000$, $\operatorname{E}[(X-$3,000)^+]=($10,000-$3,000)^2/(2\cdot$10,000)=$49,000,000/$20,000=$ $2,450.00.

Later year in nominal dollars. The trended loss is uniform on $[0,$12,000]$, and the deductible is still $3,000. Thus $\operatorname{E}[(X_1-$3,000)^+]=($12,000-$3,000)^2/(2\cdot$12,000)=$81,000,000/$24,000=$ $3,375.00.

Later year in base-year dollars. The equivalent deductible is $3,000/1.20=$2,500. The base-year severity formula gives $\operatorname{E}[(X-$2,500)^+]=($10,000-$2,500)^2/$20,000=$56,250,000/$20,000=$ $2,812.50. Re-inflating checks the nominal result: $1.20($2,812.50)=$3,375.00$.

The fixed deductible has lost value relative to severity: its base-year equivalent fell from $3,000 to $2,500. Consequently, the later expected payment increases not only in nominal dollars, but also when expressed in base-year dollars.

Common mistake and correction. A mistaken calculation would use the base-year deductible of $3,000 directly with the base-year distribution and then multiply by 1.20, obtaining $1.20($2,450)=$2,940$. This is wrong because the $3,000 later-year deductible is only $2,500 in base-year dollars. The correct calculation uses $2,500 before re-inflating, giving $3,375.

Worked Example: Expected Annual Payments with Different Year Factors

A hypothetical portfolio has base-year claim severity uniform on $[0,$20,000]$. It expects 100 claims in year 0, 80 in year 1, and 120 in year 2. The trend factors are $c_0=1$, $c_1=1.10$, and $c_2=1.21$. The deductibles are $2,000, $2,200, and $3,000, respectively. Calculate the expected total insurer payment across the three years in nominal dollars and in base-year dollars. Treat the stated claim counts as fixed.

Method. For each year, convert the deductible and severity to the same price level. For uniform severity on $[0,b]$, use $(b-a)^2/(2b)$ for the expected payment above deductible $a$. Multiply the per-claim expectation by that year's fixed claim count, then sum.

Year 0. The per-claim expectation is $($20,000-$2,000)^2/(2\cdot$20,000)=$324,000,000/$40,000=$8,100$. For 100 claims, the expected total is $100($8,100)=$ $810,000, in either price basis.

Year 1. The deductible is indexed: $2,200/1.10=$2,000 in base-year dollars. The base-year per-claim expectation is $8,100, so the nominal per-claim expectation is $1.10($8,100)=$8,910$. For 80 claims, the nominal total is $80($8,910)=$ $712,800. In base-year dollars it is $$712,800/1.10=$ $648,000.

Year 2. The base-year equivalent deductible is $3,000/1.21=$2,479.3388. Equivalently, the trended severity is uniform on $[0,$24,200]$, so the nominal per-claim expectation is $($24,200-$3,000)^2/(2\cdot$24,200)=$449,440,000/$48,400=$9,285.9504$. For 120 claims, the nominal total is $120($9,285.9504)=$1,114,314.0496$, or $1,114,314.05. Deflating gives $$1,114,314.0496/1.21=$920,920.702$, or $920,920.70 in base-year dollars.

Portfolio totals. The nominal expected total is $$810,000+$712,800+$1,114,314.0496=$2,637,114.0496$, or $2,637,114.05. The base-year-dollar total is $$810,000+$648,000+$920,920.702=$2,378,920.702$, or $2,378,920.70.

The two totals answer different questions: the nominal total measures expected payments in each year's dollars, while the base-year total puts every year's payment on a common price basis. The method avoids averaging trend factors or deductibles before applying the nonlinear payment function.

A Reliable Calculation Sequence

For a comparison involving several policy years, work year by year. The order matters because applying a deductible is nonlinear: in general, averaging loss levels or deductibles first does not preserve the expected payment.

Choose the price basis.
State whether the requested result is in each policy year's nominal dollars or in a common base-year dollar basis.

Identify each trend factor and deductible.
For year $j$, record $c_j$, the base-year severity distribution, and the deductible $d_j$ in that year's dollars.

Put the deductible in the chosen units.
For base-year calculations, use $d_j/c_j$. For nominal calculations, use the trended severity or calculate in base-year dollars and multiply by $c_j$.

Calculate expected payment separately by year.
Apply the severity model's expected-excess formula using that year's equivalent deductible.

Apply exposures and sum only after the year-specific calculations.
If the number of claims is fixed at $n_j$, the expected total for year $j$ is $n_j\operatorname{E}[Y_j^L]$. Add these totals only when they are expressed in the same price basis.

A standard ordinary-deductible loss elimination ratio can provide a check when it is already known. For a nonnegative loss with positive mean, define the loss elimination ratio at deductible $a$ as $\operatorname{E}[\min(X,a)]/\operatorname{E}[X]$. Since $X=\min(X,a)+(X-a)^+$, the expected insurer payment is $\operatorname{E}[X]$ times one minus that ratio. Across policy years, use the severity mean and deductible from a consistent price basis; do not pair a nominal deductible with a base-year severity mean.

Common Interpretation Traps

Mixing nominal and base-year dollars. A payment calculated from the base-year distribution and an unadjusted later-year deductible combines unlike units. Divide the later-year deductible by its trend factor first.
Deflating the loss but not the deductible. If working entirely with base-year severity, every deductible must also be converted to base-year dollars.
Assuming every deductible is indexed. A fixed nominal deductible does not rise with claim-cost trend. Its equivalent base-year value falls as the trend factor grows.
Comparing nominal amounts as if they measured real changes. A nominal increase can reflect general claim-cost trend even when the base-year expected payment is unchanged.
Applying a deductible after averaging across years. Calculate expected payments separately for each year's severity and deductible. The positive-part function $(X-d)^+$ is nonlinear, so an average loss or deductible cannot generally replace these separate calculations.

Key takeaway: Under proportional trend, a policy-year deductible $d_j$ is equivalent to $d_j/c_j$ in base-year dollars. Calculate the expected excess payment using matched price units, then re-inflate only when a nominal payment is required.

Check Your Understanding

For each problem, distinguish the requested price basis and apply the deductible to the matching severity distribution.

A base-year loss is uniform on $[0,$12,000]$. In a later year, $c=1.10$ and the nominal deductible is $2,200. Find the expected payment per loss in nominal dollars. (Answer: $4,583.33)
A base-year loss is exponential with mean $8,000. A later-year trend factor is $1.25$, and the nominal deductible is $5,000. Find the expected payment per loss in nominal dollars. (Answer: $6,065.31)
A base-year loss is uniform on $[0,$30,000]$. Severity trends by $1.10$, and the deductible is indexed to $3,300. Find the expected payment per loss in both nominal and base-year dollars. (Answer: $13,365.00 nominal; $12,150.00 in base-year dollars)
A base-year loss is uniform on $[0,$10,000]$. In year 0, there are 50 claims, $c_0=1$, and the deductible is $2,500. In year 1, there are 80 claims, $c_1=1.20$, and the deductible remains fixed at $2,500. Find the combined expected total in nominal dollars. (Answer: $441,458.33)

What You'll Learn