Surplus Paths and Ruin at Claim Times

Why the Shape of the Surplus Path Matters

In the classical surplus model, the insurer receives premiums continuously and pays claims at discrete arrival times. Between claims, surplus follows a straight line; at a claim, it jumps downward. This shape gives a useful way to analyze ruin: rather than searching continuously over time, we can check the surplus immediately after each claim.

That conclusion depends on the model. It uses the assumption that premium income is nonnegative and arrives continuously, with no other source of losses between claims. If surplus can fall between claims for another reason, checking only claim arrivals need not be sufficient.

Definition: In the classical surplus model, let $u\geq 0$ be initial surplus, let $c\geq 0$ be the constant premium income per unit time, and let $N(t)$ count claims by time $t$. Write $T_n$ for the time of the $n$th claim and $X_n\geq 0$ for its amount. With no investment return, the surplus at time $t$ is $$ U(t)=u+ct-\sum_{n=1}^{N(t)}X_n. $$ The surplus just before a claim at $T_n$ is $U(T_n-)$; just after it is $U(T_n)=U(T_n-)-X_n$. We use the strict ruin convention: ruin occurs when $U(t)<0$, and $\tau=\inf\{t\geq 0:U(t)<0\}$.

The classical model usually assumes a homogeneous Poisson claim count with rate $\lambda$, independent claim amounts that are identically distributed, and independence between the claim count and amounts. For the path argument below, the key requirements are simpler: claim times do not accumulate over a finite interval, claim amounts are nonnegative, and the premium rate satisfies $c\geq 0$. The Poisson and severity assumptions describe a standard stochastic model; they are not needed to establish the path shape.

Surplus Between Claim Arrivals

Suppose the $n$th claim has just been paid. Until the next claim, the count $N(t)$ does not change, so the claim total in the surplus formula stays fixed. For $T_n\leq t<T_{n+1}$, the surplus therefore satisfies

$$ U(t)=U(T_n)+c(t-T_n). $$

If the post-claim balance $U(T_n)$ is nonnegative and $c\geq 0$, then $c(t-T_n)\geq 0$. Consequently $U(t)\geq U(T_n)\geq 0$ throughout the interval. Surplus cannot first cross below zero between these two claims.

Theorem: Assume $u\geq 0$, $c\geq 0$, and that surplus changes only through continuous premium income between locally finite claim arrivals, with each claim causing a downward jump. Under the strict ruin convention, if ruin occurs, its first occurrence is at a claim arrival. Equivalently, for any $t\geq 0$, ruin has occurred by time $t$ exactly when at least one claim arrival $T_n\leq t$ has $U(T_n)<0$.

The statement concerns the first time surplus becomes negative. It does not say that surplus must stay negative after ruin: later premium income can raise it again. For detecting whether ruin has occurred, the relevant question is whether any post-claim balance has been negative.

Checking Claim-Time Balances

A direct calculation proceeds one claim at a time. First accumulate premium from the previous claim time to the next arrival. That gives the pre-claim surplus. Then subtract the new claim to get the post-claim surplus. A negative post-claim balance means ruin at that arrival; a zero balance does not under the strict convention.

Set the starting balance.
Use $U(0)=u$, and record the first claim time $T_1$.

Accumulate premium.
Before claim $n$, calculate $U(T_n-)=U(T_{n-1})+c(T_n-T_{n-1})$, with $U(T_0)=u$ and $T_0=0$.

Apply the claim jump.
Calculate $U(T_n)=U(T_n-)-X_n$. If this is negative, the first such arrival is the time of ruin.

Stop or continue.
If the post-claim balance is nonnegative, continue to the next arrival. If a finite horizon is specified, include claims occurring exactly at the horizon.

Worked Examples

Worked Example: Locating the First Ruin After Several Claims

A hypothetical insurer begins with $u=\$12$ and receives premiums at $c=\$4$ per year. Assume no investment return, a Poisson claim count with rate $\lambda=1$ claim per year, and independent claim amounts with an exponential distribution of mean $\$8$. The observed claim times and amounts are $(T_1,X_1)=(1,7)$, $(T_2,X_2)=(3,14)$, $(T_3,X_3)=(4,7)$, and $(T_4,X_4)=(4.25,2)$, with time in years and amounts in dollars. Find the first ruin time under the strict convention.

Method. Calculate each pre-claim and post-claim balance, stopping at the first negative post-claim balance.

At $T_1=1$, premium income since time zero is $4(1)=4$. Thus $U(T_1-)=12+4=16$, and after paying the first claim, $U(T_1)=16-7=9$.

From time 1 to time 3, premium income is $4(3-1)=8$. Hence $U(T_2-)=9+8=17$, and $U(T_2)=17-14=3$.

From time 3 to time 4, premium income is $4(4-3)=4$. Hence $U(T_3-)=3+4=7$, and $U(T_3)=7-7=0$. Equality to zero is not ruin under the stated convention.

From time 4 to time 4.25, premium income is $4(4.25-4)=1$. Therefore $U(T_4-)=0+1=1$, and $U(T_4)=1-2=-1$.

The first negative balance occurs at $T_4=4.25$, so $\tau=4.25$ years. The calculation also checks the zero-balance boundary at $T_3$: it is not mistaken for strict ruin. Common mistake avoided: treating a post-claim balance of exactly zero as negative.

Worked Example: Premium Earned Before the First Claim

A hypothetical insurer has $u=\$5$, $c=\$3$ per year, and no investment return. Claim arrivals follow a Poisson process with rate $\lambda=2$ per year, and claim amounts are independent exponential random variables with mean $\$4$. Suppose the first two observed claims are $(T_1,X_1)=(0.5,6)$ and $(T_2,X_2)=(1.5,4)$. Identify the first ruin time and explain why the first claim does not cause ruin.

Method. Include premium income earned before each arrival before comparing the claim with available surplus.

At $T_1=0.5$, the insurer has earned $3(0.5)=\$1.50$ in premiums. The pre-claim surplus is $U(T_1-)=5+1.50=\$6.50$. After the claim, $U(T_1)=6.50-6=\$0.50$, which is positive.

The time between the first and second claims is $1.5-0.5=1$ year, so premium income during that interval is $3(1)=\$3$. Thus $U(T_2-)=0.50+3=\$3.50$. The second claim leaves $U(T_2)=3.50-4=-\$0.50$.

The first claim does not cause ruin, and the first ruin occurs at $T_2=1.5$ years. Common mistake avoided: comparing the first claim amount of $\$6$ only with the initial $\$5$ and declaring immediate ruin. The correct comparison is with the $\$6.50$ available immediately before that claim.

Worked Example: A Claim at the End of the Horizon

A hypothetical insurer starts with $u=\$10$, earns premiums at $c=\$2.50$ per year, and has no investment return. Assume Poisson claim arrivals with rate $\lambda=0.4$ per year and independent exponential claim amounts with mean $\$6$. The observed claims are $(2,8)$, $(2.5,6)$, and $(5,9)$, where each pair gives time in years and claim amount in dollars. Determine whether ruin has occurred by time $5$, and whether it has occurred by time $4$.

Method. Calculate the post-claim balances in time order, remembering that a claim at the horizon is included when checking ruin by that time.

At $T_1=2$, premium income is $2.50(2)=\$5$, so $U(T_1-)=10+5=15$ and $U(T_1)=15-8=7$.

From time 2 to time 2.5, premium income is $2.50(0.5)=\$1.25$. Thus $U(T_2-)=7+1.25=8.25$, and $U(T_2)=8.25-6=2.25$.

From time 2.5 to time 5, premium income is $2.50(2.5)=\$6.25$. Therefore $U(T_3-)=2.25+6.25=8.50$, and $U(T_3)=8.50-9=-\$0.50$.

The first two post-claim balances are nonnegative, and there are no claims between $T_2=2.5$ and $T_3=5$. There is no ruin by time $4$. The third claim occurs exactly at time $5$ and makes surplus negative, so ruin has occurred by time $5$, with $\tau=5$. Common mistake avoided: excluding a claim at the horizon when the question asks about ruin by that time.

Why the Claim-Time Test Works

The result follows directly from the linear path between claims. At the start of any interval $[T_n,T_{n+1})$, suppose the post-claim balance is nonnegative. For any time $t$ in that interval,

U(t)=U(T_n)+c(t-T_n)\geq U(T_n)\geq 0,

because $c\geq 0$ and $t-T_n\geq 0$. Thus no time in that interval can be the first time surplus becomes negative. The only remaining possible first crossings are the downward jumps at claim arrivals. At arrival $T_n$, the jump takes surplus from $U(T_n-)$ to $U(T_n-)-X_n$; if this new value is negative, ruin occurs at $T_n$.

The explicit claim-time balance is also useful:

U(T_n)=u+cT_n-\sum_{j=1}^{n}X_j.

This expression sums all premium income earned up to $T_n$, then subtracts every claim paid by that time. It agrees with the recursive calculation, but the recursion often makes it easier to catch an arithmetic or timing error because each interval’s premium is visible.

Conventions and Limits of the Result

The strict condition $U(t)<0$ matters. If a claim reduces surplus exactly to zero, that claim does not cause ruin under this convention. If instead a problem defines ruin as $U(t)\leq 0$, equality does count; the convention must be checked before deciding a boundary case. When $u=0$, the strict convention also means the initial balance itself is not ruin. If $c>0$, surplus becomes positive immediately after time zero unless a claim arrives; a Poisson process has probability zero of an arrival exactly at time zero.

The claim-time test is specific to a surplus path that cannot decline between claims. It may fail if the model includes continuous operating expenses that exceed premium income, random investment losses between claims, or a continuous stochastic component. In those settings, surplus may cross below zero between claim arrivals, so examining only post-claim balances would miss possible ruin.

Key takeaway: In the classical model with $c\geq 0$, surplus is nondecreasing between claims. To find the first strict ruin time, accumulate premiums to each claim, subtract that claim, and stop at the first negative post-claim balance. A zero balance is not ruin under the strict convention.

Check Your Understanding

For each scenario, use the strict ruin convention and assume the classical surplus model with no investment return. Claim times and amounts are observed values; determine the requested claim-time outcome.

An insurer has $u=\$4$ and $c=\$2$ per year. Its first two claims are $(T_1,X_1)=(1,5)$ and $(T_2,X_2)=(3,7)$. At which claim does ruin first occur? (Answer: At $T_2=3$; the post-claim balances are $\$1$ and then $-\$2$.)
An insurer has $u=\$9$, $c=\$1.50$ per year, and one claim at $T_1=2$ of $\$12$. Has ruin occurred by time $4$, if there are no further claims through time $4$? (Answer: No; the post-claim balance at time 2 is $9+1.50(2)-12=0$, which is not negative, and surplus then increases.)
An insurer has $u=\$3$, $c=\$4$ per year, and claims $(0.5,3)$ and $(1.5,9)$. Find the first ruin time. (Answer: $T_2=1.5$; the first post-claim balance is $3+4(0.5)-3=2$, and the second is $2+4(1)-9=-3$.)
An insurer has $u=\$20$, $c=\$2$ per year, and claims $(1,10)$ and $(4,18)$. Has strict ruin occurred by time $4$? (Answer: No; the post-claim balances are $20+2(1)-10=12$ and $12+2(3)-18=0$. The claim at time 4 leaves zero, not a negative balance.)
An insurer has $u=\$8$, $c=\$2$ per year, and claims $(2,7)$, $(3,6)$, and $(3.5,3)$. Find the first ruin time. (Answer: $T_3=3.5$; the successive post-claim balances are $8+2(2)-7=5$, $5+2(1)-6=1$, and $1+2(0.5)-3=-1$.)

What You'll Learn