Transition Benefits in the Multi-State Reserve Equation

Why Transition Benefits Need Their Own Term

In a multi-state model, a policy can move from one state to another, and a payment may be made at the instant of that move. The reserve after the move is generally different from the reserve before it: the destination state can have a different pattern of future benefits and premiums. The source-state reserve equation must therefore account for both the transition payment and the destination-state reserve.

The key comparison is between the source-state reserve immediately before the transition and the insurer’s total liability immediately after it. That post-transition liability includes any payment made at the transition and the reserve for future obligations in the destination state. The transition intensity weights this difference to give a contribution per unit time to the source-state reserve derivative.

Definition: Let $V_i(t)$ be the prospective reserve at time $t$, conditional on the contract being in state $i$. For distinct states $i$ and $j$, let $\mu_{ij}(t)$ be the transition intensity from $i$ to $j$: $$ \mu_{ij}(t)=\lim_{h\downarrow 0}\frac{P(\text{state at }t+h\text{ is }j\mid\text{state at }t\text{ is }i)}{h}. $$ Let $b_{ij}(t)$ be the amount paid by the insurer at that transition, with an insurer payment positive and an insurer receipt negative. Thus, a receipt of $R$ has $b_{ij}=-R$. The destination-state reserve $V_j(t)$ is the reserve for future obligations after the transition payment has been accounted for.

The intensity has units of inverse time, while $V_i$, $V_j$, and $b_{ij}$ have units of money. Consequently, $\mu_{ij}(V_i-V_j-b_{ij})$ has units of money per unit time, as a term in a reserve derivative must.

The Transition-Benefit Term

Use the standard continuous-time multi-state reserve setup: transitions follow specified intensities; over a sufficiently short interval, the probability of one transition is its intensity times the interval length to first order; and simultaneous transitions have probability of smaller order. Assume the force of interest is $\delta$. Write $c_i(t)$ for the continuous net outgo rate in state $i$, positive when it is an outgo to the insurer and negative when it is an inflow. This notation isolates the transition term; the treatment of state-specific cash-flow rates is not the focus here.

Theorem: Under these conventions, the reserve in state $i$ satisfies $$ \frac{dV_i(t)}{dt} = \delta V_i(t)-c_i(t) + \sum_{j\ne i}\mu_{ij}(t)\bigl(V_i(t)-V_j(t)-b_{ij}(t)\bigr). $$ For the transition $i\to j$, its contribution is $$ \mu_{ij}(t)\bigl(V_i(t)-V_j(t)-b_{ij}(t)\bigr). $$

The benefit sign follows directly from the convention. A payment $b_{ij}>0$ increases the insurer’s outgo when the transition occurs, so it is subtracted inside the parentheses. A receipt has $b_{ij}<0$; subtracting that negative amount increases the transition contribution. The destination reserve is also subtracted because it is the reserve needed for the future obligations after the transition.

Why the Formula Holds

The short-interval reserve recursion makes the signs explicit. Starting in state $i$ at time $t$, during an interval of length $h$, the probability of transition $i\to j$ is $\mu_{ij}(t)h+o(h)$. The probability of no transition is $1-h\sum_{j\ne i}\mu_{ij}(t)+o(h)$. To first order, the insurer’s outgo during the interval is the continuous outgo $c_i(t)h$, plus the transition payment and continuation reserve if a transition occurs. Discounting the end-of-interval value at force $\delta$ gives

V_i(t)=e^{-\delta h}\left[ c_i(t)h+ \sum_{j\ne i}\mu_{ij}(t)h\bigl(b_{ij}(t)+V_j(t+h)\bigr) +\left(1-h\sum_{j\ne i}\mu_{ij}(t)\right)V_i(t+h) \right]+o(h).

Use $e^{-\delta h}=1-\delta h+o(h)$, $V_i(t+h)=V_i(t)+hV_i'(t)+o(h)$, and $V_j(t+h)=V_j(t)+o(1)$. After substituting and cancelling $V_i(t)$, divide by $h$ and let $h\downarrow0$. The resulting equation is

0=c_i(t)+ \sum_{j\ne i}\mu_{ij}(t)\bigl(b_{ij}(t)+V_j(t)-V_i(t)\bigr) +V_i'(t)-\delta V_i(t),

which rearranges to the stated multi-state reserve equation. In particular, the transition term is not just an expected benefit rate: it also reflects the change from the source reserve to the destination reserve.

Calculating Contributions From Several Transitions

Each possible destination is handled separately using its own intensity, destination reserve, and transition payment. Then add the contributions. The sum is over transitions out of the source state; it does not include a self-transition.

Worked Example: Two Possible Transitions From One State

A hypothetical contract is in state $A$. Its reserve is $V_A=\$1{,}200$. Transitions to $B$ and $C$ have intensities $0.06$ and $0.02$ per year, respectively. The destination reserves are $V_B=\$500$ and $V_C=\$100$. The insurer pays $\$100$ on transition to $B$ and $\$900$ on transition to $C$. Find the total transition contribution to $dV_A/dt$.

Method. For each transition, calculate $\mu_{Aj}(V_A-V_j-b_{Aj})$. Intensities are annual, so each result is in dollars per year.

For $A\to B$, $b_{AB}=100$. The reserve difference after allowing for the payment is $1{,}200-500-100=600$. Therefore the contribution is $0.06(600)=\$36$ per year.

For $A\to C$, $b_{AC}=900$. The corresponding difference is $1{,}200-100-900=200$. Therefore the contribution is $0.02(200)=\$4$ per year.

Adding the two contributions gives $36+4=\$40$ per year. Check: the total outgo at either transition, including the destination reserve, is respectively $\$600$ and $\$1{,}000$, both below the current reserve of $\$1{,}200$. The positive contributions are consistent with that comparison. The common mistake avoided here is to omit the destination reserve or to use one transition’s intensity for both destinations.

Receipts and the Reserve Derivative

A transition receipt is not a negative reserve. It is a cash flow with a negative value of $b_{ij}$. Keep the same formula and substitute the signed amount. The next example combines a receipt with the other terms in the source-state equation.

Worked Example: A Receipt and the Full Reserve Equation

A hypothetical policy is in state $M$, with $V_M=\$900$ and $V_N=\$400$. The transition intensity from $M$ to $N$ is $0.05$ per year. At that transition, the insurer receives $\$80$. The force of interest is $\delta=0.03$ per year, and the continuous net outgo rate in state $M$ is $c_M=\$12$ per year. Find $dV_M/dt$.

Method. First write the signed transition amount. Since the insurer receives $\$80$, $b_{MN}=-80$. The transition contribution is $0.05(900-400-(-80))$.

The reserve difference inside the parentheses is $900-400+80=580$. Thus the transition contribution is $0.05(580)=\$29$ per year. The interest term is $\delta V_M=0.03(900)=\$27$ per year, and the continuous net outgo term is $-c_M=-\$12$ per year.

Therefore $dV_M/dt=27-12+29=\$44$ per year. Check: treating the receipt as an outgo would give $0.05(900-400-80)=\$21$ per year for the transition term, which is incorrect. With the receipt, the insurer collects $\$80$ as well as moving to a lower reserve, so the correct contribution is $\$29$ per year.

Using the Equation to Find an Unknown Transition Benefit

The equation can also be rearranged when the reserve derivative is known and a transition benefit is unknown. This is useful for checking whether a proposed benefit amount is consistent with a specified reserve path. Keep the sign convention explicit throughout: solve for $b_{ij}$ as an insurer outgo, so a positive answer is a payment and a negative answer is a receipt.

Worked Example: Solving for a Transition Payment

A hypothetical contract in state $U$ has $V_U=\$1{,}000$. It can move to $V$ with intensity $0.05$ per year or to $W$ with intensity $0.02$ per year. The destination reserves are $V_V=\$600$ and $V_W=\$100$. The insurer pays an unknown amount $b_{UV}$ on transition to $V$, and receives $\$50$ on transition to $W$. Let $\delta=0.03$, $c_U=\$10$ per year, and suppose $dV_U/dt=\$44$ per year. Find $b_{UV}$.

Method. The receipt on transition to $W$ is $b_{UW}=-50$. The interest and continuous-outgo terms combine to $0.03(1{,}000)-10=30-10=\$20$ per year.

The contribution from $U\to W$ is $0.02(1{,}000-100-(-50))=0.02(950)=\$19$ per year. The $U\to V$ contribution is $0.05(1{,}000-600-b_{UV})=0.05(400-b_{UV})$.

Substitute these terms into the equation: $44=20+19+0.05(400-b_{UV})$. Subtract $39$ from both sides to obtain $5=0.05(400-b_{UV})$. Divide by $0.05$: $100=400-b_{UV}$, so $b_{UV}=\$300$.

Check. With a $\$300$ payment, the $U\to V$ contribution is $0.05(1{,}000-600-300)=\$5$ per year. The full derivative is $20+19+5=\$44$ per year, as specified. The common mistake avoided is to treat the $\$50$ receipt on the other transition as a positive outgo; its signed amount is $-50$.

Common Sign and Interpretation Traps

The safest approach is to write the amount $b_{ij}$ from the insurer’s perspective before calculating. The following comparison summarizes the effect of the sign.

Cash flow at transition $i\to j$	Value of $b_{ij}$	Term inside the intensity-weighted contribution
Insurer pays amount $B$	$B$	$V_i-V_j-B$
Insurer receives amount $R$	$-R$	$V_i-V_j+R$
No transition payment	$0$	$V_i-V_j$

A common wrong calculation is to insert a receipt as though it were a payment. For example, with $V_i=\$900$, $V_j=\$400$, intensity $0.05$, and an insurer receipt of $\$80$, the incorrect calculation is $0.05(900-400-80)=\$21$ per year. The correction is to set $b_{ij}=-80$, giving $0.05(900-400-(-80))=\$29$ per year.

Another common error is to use only the transition payment and ignore $V_j$. The transition does not end the contract merely because it changes state: future obligations in the destination state remain represented by $V_j$. Conversely, including $V_j$ in the transition term does not mean the transition payment has already been included in that reserve; under the convention used here, the payment at the transition and the destination reserve are separate components.

Key takeaway: For every transition out of state $i$, use the same signed formula: intensity times source reserve minus destination reserve minus the signed insurer payment. Sum those contributions, then combine them with interest and continuous net outgo to obtain the source-state reserve derivative.

Check Your Understanding

For each problem, use the insurer-outgo convention for transition benefits. Intensities and reserve derivatives are per year.

From state $A$, a transition to $B$ has intensity $0.03$. If $V_A=\$800$, $V_B=\$300$, and the insurer pays $\$100$ on the transition, find its contribution to $dV_A/dt$. (Answer: $\$12$ per year)
From state $C$, transitions to $D$ and $E$ have intensities $0.04$ and $0.01$. The reserves are $V_C=\$900$, $V_D=\$400$, and $V_E=\$0$. The insurer pays $\$200$ on the transition to $D$ and receives $\$50$ on the transition to $E$. Find the total transition contribution. (Answer: $\$21.50$ per year)
In state $H$, $\delta=0.04$, $V_H=\$500$, and $c_H=\$8$ per year. There is one transition to $J$, with intensity $0.06$, $V_J=\$250$, and an insurer payment of $\$100$. Find $dV_H/dt$. (Answer: $\$21$ per year)
In state $K$, $V_K=\$1{,}200$. Transitions to $L$ and $M$ have intensities $0.05$ and $0.02$, with $V_L=\$800$ and $V_M=\$100$. The insurer pays an unknown amount on transition to $L$ and receives $\$100$ on transition to $M$. If $\delta=0.02$, $c_K=\$5$ per year, and $dV_K/dt=\$30$ per year, find the payment on transition to $L$. (Answer: $\$660$)
Suppose a transition has intensity $0.04$, source reserve $\$1{,}000$, destination reserve $\$700$, and the insurer receives $\$50$. Find the transition contribution and state the signed transition payment used. (Answer: $\$14$ per year; $b_{ij}=-\$50$)

Cash flow at transition \(i\to j\)	Value of \(b_{ij}\)	Term inside the intensity-weighted contribution
Insurer pays amount \(B\)	\(B\)	\(V_i-V_j-B\)
Insurer receives amount \(R\)	\(-R\)	\(V_i-V_j+R\)
No transition payment	\(0\)	\(V_i-V_j\)

What You'll Learn