probability of ruin

Even a well-run insurer faces a nightmare scenario: a bad enough run of claims that the money runs out and the surplus drops below zero. The probability of ruin is simply the chance that this ever happens — the likelihood that, somewhere along the jagged surplus line, the balance dips below zero and the insurer is technically insolvent. It is the headline number that risk theory exists to estimate.

Precisely, ruin is the event that the surplus U(t) becomes negative for some time t, and the probability of ruin starting from initial surplus u is written psi(u) (the Greek letter psi). Two features stand out. First, psi(u) decreases as u rises: more starting capital means a smaller chance of ever going broke. Second, in the classical model with a positive loading, psi(u) is strictly between 0 and 1 — ruin is possible but not certain — and it shrinks roughly exponentially as u grows. For example, you might find psi(0) = 0.6 with no buffer, falling to psi(2,000,000) = 0.01 once a healthy cushion is in place.

The probability of ruin is the bridge from pricing to solvency: it turns 'are our premiums and capital adequate?' into a single number you can target. Regulators and risk managers often invert it — pick an acceptable ruin probability such as 0.5 percent over a year and solve for the capital u that achieves it. Two honest caveats. The classical psi(u) is usually an infinite-time probability (ruin ever, over an unbounded horizon), which is more pessimistic than the one-year view real firms care about. And it is a model output: it inherits every simplification (no investment income, idealized claim distribution) and is only as trustworthy as those assumptions.