Skip to content

Economic Modelling · Ruin theory

Exact Ruin Probabilities and Special Cases for Exponential Claims

Updated 11 October 2026 · Fact-checked

Ruin probability ψ(u) is the chance the surplus ever falls below zero in infinite time, starting from surplus u. For exponential claims with mean μ and relative loading θ, ψ(u) = (1 ÷ (1+θ)) × exp(−θu ÷ ((1+θ)μ)). At u = 0 it equals 1 ÷ (1+θ).

Understand Exact Ruin Probabilities and Special Cases

In the compound Poisson surplus model, U(t) = u + ct − S(t). Here u is the initial surplus, c is the premium rate and S(t) is total claims up to time t. Claims arrive at Poisson rate λ. Individual claims X have mean μ. The premium is c = (1+θ)λμ, where θ > 0 is the relative security loading. Ruin probability in infinite time is ψ(u) = P(U(t) < 0 for some t > 0 | U(0) = u).

Exact formulas for ψ(u) are rare. The main exact case is exponential claims, where X has density f(x) = (1/μ)e^(−x/μ). The lack-of-memory property makes the maths tractable. When the surplus first dips below its previous low, the size of the dip (the deficit) is again exponential with mean μ. This gives a clean closed form.

At u = 0 something simple happens for any claim distribution with mean μ. The ruin probability is ψ(0) = 1 ÷ (1+θ). It does not depend on the shape of the claim distribution, only on the loading. For exponential claims the general formula gives the same value, which is a useful check.

The Cramér-Lundberg approximation applies when the adjustment coefficient R exists. It says ψ(u) ≈ C e^(−Ru) for large u, where C is a constant. Lundberg's inequality only gives the bound ψ(u) ≤ e^(−Ru). The approximation gives a sharper estimate of size. For exponential claims the approximation is in fact exact, with C = 1 ÷ (1+θ) and R = θ ÷ ((1+θ)μ).

Key rules to remember

Premium with loading
c = (1+θ)λμ
θ is the relative security loading. Net claims rate is λμ. You need θ > 0 for ruin probability to be below 1.
Ruin probability at zero surplus
ψ(0) = 1 ÷ (1+θ)
Holds for the compound Poisson model with any claim size distribution that has finite mean, in infinite time.
Exponential claims, exact ruin probability
ψ(u) = (1 ÷ (1+θ)) × exp(−θu ÷ ((1+θ)μ))
Claims exponential with mean μ. Equivalent form: ψ(u) = (λμ ÷ c) × e^(−Ru).
Adjustment coefficient, exponential claims
R = θ ÷ ((1+θ)μ)
With rate parameter β = 1/μ, this is R = β − λ/c.
General adjustment coefficient equation
λ + cR = λ M_X(R), R > 0
M_X(R) = E[e^(RX)] is the moment generating function of a claim. Take the positive root.
Lundberg's inequality
ψ(u) ≤ e^(−Ru)
A bound, valid for u ≥ 0 when R exists.
Cramér-Lundberg approximation
ψ(u) ≈ C e^(−Ru)
Asymptotic for large u. For exponential claims C = 1 ÷ (1+θ) and it is exact for all u ≥ 0.

How to solve Exact Ruin Probabilities and Special Cases questions

Use this method for any question on exact or approximate infinite-time ruin probability in the compound Poisson model.

  1. 1Write down the model: λ, claim distribution, mean μ, premium c and initial surplus u.
  2. 2Find θ from c = (1+θ)λμ, so θ = c ÷ (λμ) − 1. Check θ > 0.
  3. 3If u = 0, use ψ(0) = 1 ÷ (1+θ) straight away. No claim distribution detail is needed.
  4. 4If claims are exponential, compute R = θ ÷ ((1+θ)μ) and then ψ(u) = (1 ÷ (1+θ)) e^(−Ru).
  5. 5For other claim distributions, solve λ + cR = λ M_X(R) for the positive root R. Use the bound e^(−Ru) or the approximation C e^(−Ru) as the question asks.
  6. 6Substitute u, keep units consistent (same time unit for λ and c, same money unit for μ and u), and calculate.
  7. 7Sanity check: the answer must lie between 0 and 1, fall as u rises, and rise as θ falls.

Quickest way: Exponential claims in three lines

When to use it: When the question says claims are exponential and gives λ, μ and c (or θ), and asks for ψ(u) or ψ(0).

  1. Get θ = c ÷ (λμ) − 1.
  2. Compute 1 ÷ (1+θ). This is ψ(0) and also the constant C.
  3. Compute R = θ ÷ ((1+θ)μ), then ψ(u) = (1 ÷ (1+θ)) e^(−Ru).

Common mistakes in Exact Ruin Probabilities and Special Cases

  • Using c as the loading, for example writing ψ(0) = 1 ÷ c.

    Students mix up premium rate and the loading factor.

    Fix: Always compute θ = c ÷ (λμ) − 1 first. ψ(0) = 1 ÷ (1+θ) = λμ ÷ c.

  • Writing R = θ ÷ μ for exponential claims.

    The factor (1+θ) in the denominator is forgotten.

    Fix: Remember R = θ ÷ ((1+θ)μ). Check by R = 1/μ − λ/c.

  • Dropping the constant 1 ÷ (1+θ) and quoting e^(−Ru) as the exact probability.

    Lundberg's bound is confused with the exact answer.

    Fix: e^(−Ru) is only an upper bound. The exact exponential result has the factor 1 ÷ (1+θ) in front.

  • Saying ψ(0) depends on the claim distribution, or that ψ(0) = 1 when u = 0.

    Students think zero surplus means certain ruin.

    Fix: With positive loading, ψ(0) = 1 ÷ (1+θ) < 1 for any claim distribution with finite mean.

  • Mixing time units, such as λ per month with c per year.

    Data are given in different units in the question.

    Fix: Convert λ and c to the same time unit before finding θ.

  • Stating the Cramér-Lundberg approximation without conditions.

    Memorising the formula without its requirements.

    Fix: State that it needs the adjustment coefficient R to exist and is an approximation for large u, except for exponential claims where it is exact.

Worked examples

Example 1

Claims arrive as a Poisson process with rate 5 per year. Claim sizes are exponential with mean ₹40,000. Premiums are received continuously at ₹2,80,000 per year. Find (a) the relative security loading, (b) the adjustment coefficient, (c) the probability of ultimate ruin when initial surplus is zero, and (d) when initial surplus is ₹1,00,000.

Show the solution
  1. Expected claims per year = λμ = 5 × 40,000 = ₹2,00,000.
  2. (a) θ = 2,80,000 ÷ 2,00,000 − 1 = 1.4 − 1 = 0.4.
  3. (b) R = θ ÷ ((1+θ)μ) = 0.4 ÷ (1.4 × 40,000) = 0.4 ÷ 56,000 = 7.1429 × 10⁻⁶ per rupee.
  4. (c) ψ(0) = 1 ÷ 1.4 = 0.7143.
  5. (d) Ru = 7.1429 × 10⁻⁶ × 1,00,000 = 0.71429. e^(−0.71429) = 0.4895.
  6. ψ(1,00,000) = 0.7143 × 0.4895 = 0.3497.

Answer: θ = 0.4; R ≈ 7.14 × 10⁻⁶ per rupee; ψ(0) ≈ 0.714; ψ(₹1,00,000) ≈ 0.350.

Example 2

In a compound Poisson surplus model with exponential claims of mean 10, the relative security loading is 25%. Find the initial surplus u needed so that the ultimate ruin probability is 5%.

Show the solution
  1. θ = 0.25, so 1+θ = 1.25 and μ = 10.
  2. R = 0.25 ÷ (1.25 × 10) = 0.25 ÷ 12.5 = 0.02.
  3. ψ(u) = (1 ÷ 1.25) e^(−0.02u) = 0.8 e^(−0.02u).
  4. Set 0.8 e^(−0.02u) = 0.05, so e^(−0.02u) = 0.0625.
  5. Take logs: −0.02u = ln(0.0625) = −2.77259.
  6. u = 2.77259 ÷ 0.02 = 138.63.

Answer: u ≈ 138.6 (in the same units as the claim mean), about 13.9 times the mean claim.

Exam tips

  • For exponential claims, learn ψ(u) = (1 ÷ (1+θ)) e^(−Ru) and R = θ ÷ ((1+θ)μ). Examiners often ask you to derive R first and then ψ(u).
  • Check the answer at u = 0. If your formula does not give 1 ÷ (1+θ), something is wrong.
  • In written answers, state the assumptions: compound Poisson claims, premium received continuously at constant rate, θ > 0, infinite time horizon.
  • In multiple-choice questions, look for the trap of quoting Lundberg's bound as the exact value. Compare the constant in front of the exponential.
  • For computer-based work, set up θ, R and ψ(u) as named variables so you can reuse them for different u values and show each step.

Practice questions from Ruin theory

Exact Ruin Probabilities and Special Cases: frequently asked questions

What is the ruin probability formula for exponential claims?

For claims exponential with mean μ and relative loading θ, ψ(u) = (1 ÷ (1+θ)) exp(−θu ÷ ((1+θ)μ)). It holds for infinite time in the compound Poisson model. Here u ≥ 0 is the initial surplus.

How do I find the ruin probability when initial surplus is zero?

Use ψ(0) = 1 ÷ (1+θ). It works for any claim distribution with finite mean in the compound Poisson model. Find θ from c = (1+θ)λμ.

What is the difference between Lundberg's inequality and the Cramér-Lundberg approximation?

Lundberg's inequality gives an upper bound, ψ(u) ≤ e^(−Ru). The Cramér-Lundberg approximation gives ψ(u) ≈ C e^(−Ru) for large u, with a constant C. The approximation is closer to the true value than the bound.

Is the Cramér-Lundberg approximation exact for exponential claims?

Yes. For exponential claims, C = 1 ÷ (1+θ) and R = θ ÷ ((1+θ)μ), and the formula holds exactly for every u ≥ 0. For other claim distributions it is only an approximation.