Tutorials › Exam FAM - Fundamentals of Actuarial Mathematics › Recover an Unknown Contract Parameter from Premium and Reserve Data

Exam FAM · Life Contract Calculation Chains · Tutorial 919 of 988

Recover an Unknown Contract Parameter from Premium and Reserve Data

Learn to solve for an unknown contract quantity from premium and reserve data using correctly timed annual recursions and insurance identities.

Advanced 10 min read

What You'll Learn

  • Set up the annual reserve recursion with premiums and benefits at the stated times.
  • Rearrange the recursion to recover an unknown benefit amount.
  • Infer an effective annual interest rate from one year of reserve data.
  • Solve for a one-year death probability and check its admissible range.
  • Use whole-life insurance-annuity identities as an independent contract-value check.
  • Detect inconsistent or insufficient premium-and-reserve data.

Why Premium and Reserve Data Can Reveal Contract Parameters

A premium and a reserve are linked by the contract’s cash flows, interest accumulation, and mortality. Consequently, observed values at consecutive durations can reveal a quantity that was not reported directly, such as the death benefit, the effective annual interest rate, or the probability of death during a year. The key is to preserve the timing of each cash flow when translating the data into an equation.

This tutorial uses annual, discrete contracts. In each one-year interval, the premium is paid at the beginning of the year if the insured is alive, a death benefit is paid at the end of the year if death occurs, and the next reserve applies at the beginning of the following year if the insured survives. These timing assumptions are essential; the same equation does not automatically apply to continuous benefits, midyear premiums, or contracts with additional payments.

Definition: Let \(V_k\) be the reserve immediately before the premium due at time \(k\), conditional on the insured being alive then. Let \(P\) be the annual premium, \(B\) the death benefit payable at the end of the year of death, \(i\) the effective annual interest rate, \({}_1q_{x+k}\) the probability of death during the year, and \({}_1p_{x+k}=1-{}_1q_{x+k}\) the probability of survival. For a net-premium contract with no other cash flows, the annual reserve recursion is $$ (V_k+P)(1+i)={}_1q_{x+k}B+{}_1p_{x+k}V_{k+1}. $$

The equation is conditional on being alive at time \(k\). The reserve and premium available at that time accumulate with interest for one year. At year-end, the accumulated amount must cover either the death benefit, if death occurred, or the reserve for the surviving policy, if survival occurred. The right side is therefore a probability-weighted amount, not the sum of a benefit and a reserve paid in every outcome.

Deriving and Rearranging the Annual Recursion

The recursion follows from equating the value accumulated over the year to the expected amount needed at its end. The two possible year-end outcomes are mutually exclusive and exhaustive: death with probability \({}_1q_{x+k}\), or survival with probability \({}_1p_{x+k}\). Multiplying each conditional payment by its probability gives the expected year-end liability. Equating this liability to the accumulated reserve and premium gives the displayed formula.

Depending on which quantity is unknown, rearrange the same equation rather than selecting a new formula. For a death benefit, an interest rate, and a mortality probability, respectively, the useful rearrangements are:

$$ B=\frac{(V_k+P)(1+i)-{}_1p_{x+k}V_{k+1}}{{}_1q_{x+k}}, \qquad i=\frac{{}_1q_{x+k}B+{}_1p_{x+k}V_{k+1}}{V_k+P}-1, \qquad {}_1q_{x+k}=\frac{(V_k+P)(1+i)-V_{k+1}}{B-V_{k+1}}. $$

Each rearrangement has a condition. To infer \(B\), the death probability must be nonzero. To infer \(i\), \(V_k+P\) must be nonzero. To infer mortality, \(B\) must differ from \(V_{k+1}\); otherwise the equation does not distinguish death from survival. A calculated mortality probability must also lie between zero and one, inclusive. These conditions help separate an algebraic answer from an identifiable, admissible contract parameter.

Worked Example: Recovering a Death Benefit from Two Reserves

A hypothetical annual term policy has a level net premium of \(P=\$120\), and the effective annual interest rate is \(i=0.05\). At duration 0, the reserve before the premium is \(V_0=\$400\); at duration 1, it is \(V_1=\$440\). The probability of death during the first year is \(0.10\). Find the level death benefit, then use the next year’s reserve data as a check. For the second year, the death probability is \(0.20\), and the reserve before the next premium is \(V_2=\$360\).

Method. Apply the recursion first to the interval from time 0 to time 1 and solve for \(B\). The first-year survival probability is \(1-0.10=0.90\). The accumulated reserve and premium are

$$ (V_0+P)(1+i)=(\$400+\$120)(1.05)=\$520(1.05)=\$546. $$

The expected year-end amount is \(0.10B+0.90(\$440)\). Thus,

$$ \$546=0.10B+\$396, \qquad 0.10B=\$150, \qquad B=\$1{,}500. $$

Check using the second interval. The accumulated amount is \((\$440+\$120)(1.05)=\$560(1.05)=\$588\). The year-end expected amount using the recovered benefit is \(0.20(\$1{,}500)+0.80(\$360)=\$300+\$288=\$588\). The second interval agrees, so the recovered benefit is consistent with both reserve observations.

Common mistake avoided. Subtracting the survival reserve from the accumulated amount and dividing by the death probability is valid only after accounting for the probability of survival. Here the survival contribution is \(0.90(\$440)\), not \(\$440\). Ignoring that probability would give a different and incorrect benefit.

Inferring Interest or Mortality

The same equation can identify a rate or a mortality quantity when the other inputs are known. For interest, the cash-flow amounts and probabilities determine the required accumulation factor. For mortality, the year-end expected amount is a weighted average of the death benefit and the next reserve. If those two amounts differ, the weight on the death benefit identifies the death probability.

Worked Example: Inferring the Effective Annual Interest Rate

A hypothetical one-year segment of a life contract has a death benefit of \(B=\$1{,}000\). At the start of the year, the reserve is \(V_k=\$600\) and a premium of \(P=\$200\) is paid. The one-year death probability is \(0.20\), and the next reserve is \(V_{k+1}=\$800\). Find the effective annual interest rate.

Method. The survival probability is \(1-0.20=0.80\). The expected year-end amount is

$$ 0.20(\$1{,}000)+0.80(\$800)=\$200+\$640=\$840. $$

The starting reserve plus premium is \(\$600+\$200=\$800\). Therefore the accumulation factor must satisfy

$$ \$800(1+i)=\$840, \qquad 1+i=\frac{\$840}{\$800}=1.05, \qquad i=0.05. $$

The inferred effective annual interest rate is \(5\%\). Substitution checks the result: \((\$600+\$200)(1.05)=\$840\), matching the expected year-end amount.

Common mistake avoided. The reserve alone is not the amount invested for the year because the premium is paid at the beginning of the year. Using \(\$600\) instead of \(\$800\) as the accumulation base would infer the wrong rate.

Worked Example: Inferring a One-Year Death Probability

For a hypothetical annual policy, \(V_k=\$500\), \(P=\$100\), and \(i=0.05\). The death benefit is \(B=\$1{,}430\), and the reserve at the next duration is \(V_{k+1}=\$430\). Find the one-year death probability.

Method. The accumulated reserve and premium are \((\$500+\$100)(1.05)=\$600(1.05)=\$630\). Write the year-end expected amount as the death probability times the death benefit plus the survival probability times the next reserve:

$$ \$630=q(\$1{,}430)+(1-q)(\$430) =\$430+q(\$1{,}000). $$

Subtracting \(\$430\) from both sides gives \(q(\$1{,}000)=\$200\), so \(q=0.20\). The implied survival probability is \(1-q=0.80\). Checking, \(0.20(\$1{,}430)+0.80(\$430)=\$286+\$344=\$630\), as required.

The benefit exceeds the next reserve by \(\$1{,}000\), so the observed expected amount identifies a unique mortality probability. The answer is admissible because it lies between zero and one.

Common mistake avoided. Dividing the difference \(\$630-\$430\) by the death benefit \(\$1{,}430\) would be incorrect. The probability weights a change from the survival amount \(\$430\) to the death amount \(\$1{,}430\), a difference of \(\$1{,}000\), not the full death benefit.

Using Insurance Identities as an Independent Check

For a whole-life contract with annual premiums payable in advance while alive and a unit benefit payable at the end of the year of death, the standard discrete insurance-annuity identity is

$$ A_x+d\ddot{a}_x=1, \qquad d=\frac{i}{1+i}. $$

To see why, suppose death occurs during policy year \(K+1\). The present value of the unit death benefit is \(v^{K+1}\), while the present value of premiums is \(1+v+\cdots+v^K\). Since \(d=1-v\),

$$ d(1+v+\cdots+v^K)=1-v^{K+1}. $$

Taking expectations over the future lifetime gives \(d\ddot{a}_x=1-A_x\), which is the stated identity. Under the net-premium equivalence principle, a whole-life policy with level benefit \(B\) and level annual premium \(P\) therefore satisfies \(P\ddot{a}_x=BA_x\). At duration \(k\), its prospective net reserve is \(V_k=BA_{x+k}-P\ddot{a}_{x+k}\).

These identities provide a second route when the relevant insurance and annuity values are supplied or can be calculated from the stated mortality and interest basis. For example, if \(A_{x+k}\ne 0\), the reserve identity can be rearranged to \(B=(V_k+P\ddot{a}_{x+k})/A_{x+k}\). A candidate value obtained this way should agree with the value implied by the reserve recursion when both use the same benefit definition, premium pattern, and timing.

Consistency, Identifiability, and Exam Traps

A set of observations should not be treated as automatically consistent just because one equation produces a numerical answer. When several durations are supplied, each applicable recursion should hold. A disagreement can indicate inconsistent data, a mistaken timing interpretation, or a model that omits a contract cash flow. If the data are insufficient, the correct conclusion may be that the parameter cannot be uniquely determined.

  • Keep the time points explicit. \(V_k\) is before the premium at time \(k\), while \(V_{k+1}\) is before the next premium. Moving either reserve to a different point in the year changes the equation.
  • Use conditional probabilities for that policy year. The recursion uses the mortality probability from age \(x+k\) over one year, not a probability over the full contract term.
  • Respect cash-flow timing. A premium paid at the beginning of the year earns a full year of interest in this model. A benefit paid at the end of the year is not discounted again in the recursion.
  • Check parameter bounds. An inferred effective annual interest rate should satisfy the model’s stated restrictions, and an inferred death probability must be in \([0,1]\).
  • Check whether the parameter is identifiable. If \(B=V_{k+1}\), the mortality equation has no information about \(q\). If \(V_k+P=0\), the displayed interest rearrangement cannot determine \(i\).
  • Do not apply the annual recursion unchanged to a different contract design. Expenses, multiple benefits, survival benefits, or premiums paid at another time require the corresponding cash flows to be included with their correct timing.

For a reliable calculation, first state the one-year model and cash-flow timing, then write the recursion, isolate the requested quantity, and finally substitute the answer back into the original equation. This last check catches algebra errors and many misreadings of reserve data. The sequence is aligned with the SOA Exam FAM (Fundamentals of Actuarial Mathematics) syllabus.

Check Your Understanding

For each problem, use the stated annual timing and check that the inferred quantity is consistent with the recursion.

  1. A hypothetical policy has \(i=0.04\), \(P=\$75\), \(V_k=\$250\), \(V_{k+1}=\$270\), and \({}_1q_{x+k}=0.10\). Find the death benefit. (Answer: \(\$950\))
  2. A one-year contract segment has \(B=\$1{,}200\), \({}_1q_{x+k}=0.25\), \(V_k=\$400\), \(P=\$100\), and \(V_{k+1}=\$600\). Find \(i\). (Answer: \(0.500\), or \(50.0\%\))
  3. Given \(i=0.03\), \(V_k=\$700\), \(P=\$100\), \(B=\$1{,}600\), and \(V_{k+1}=\$400\), find \({}_1q_{x+k}\). Give the probability to four decimals. (Answer: \(0.3533\))
  4. For a hypothetical policy segment, \(i=0.04\), \(P=\$90\), \(V_k=\$510\), \(B=\$1{,}200\), and \({}_1q_{x+k}=0.15\). Find \(V_{k+1}\), rounding only the final amount to the nearest cent. (Answer: \(\$522.35\))
  5. Data report \(i=0.05\), \(P=\$100\), \(V_k=\$200\), \(V_{k+1}=\$300\), \({}_1q_{x+k}=0.20\), and \(B=\$1{,}000\). Do these values satisfy the recursion? (Answer: No; the accumulated amount is \(\$315\), while the expected year-end amount is \(\$440\), a discrepancy of \(\$125\).)
Clinical Biostats