expected loss ratio method
Suppose you have written a brand-new line of business this year. There is almost no loss data yet — the year is too young, the triangle is nearly empty, and the chain ladder would be wildly unstable. So you fall back on a different question: how much premium did we collect, and what fraction of premium do we expect to pay out as losses for this kind of business? Multiply the two and you have an estimate of ultimate losses that does not depend on the (nonexistent) development data at all. That is the expected loss ratio method.
Precisely, the expected loss ratio (ELR) method estimates ultimate losses as earned premium multiplied by an assumed loss ratio, chosen in advance from pricing studies, industry data, or the company's own history on similar business. The reserve is then that estimated ultimate minus losses paid (or reported) to date. For example, if a year earned 10 million of premium and the expected loss ratio is 65 percent, the method sets ultimate losses at 6.5 million regardless of how the early claims happen to be developing; if 1 million has been paid, the reserve is 5.5 million. Crucially, the ELR estimate completely ignores the actual emerging loss experience — it is an outside-in answer.
The ELR method matters as the natural reserve for the greenest accident years and for new lines where development data is too thin to trust. Its strength — stability — is also its weakness: because it ignores actual claims, a year that is clearly running hot or cold gets the same answer either way, which can be dangerously stubborn if the chosen loss ratio was wrong. This tension is why the Bornhuetter-Ferguson method was invented: it blends the ELR's stable expectation with the chain ladder's responsiveness to real data, weighting toward expectation when the year is young and toward data as it matures.
A new cyber-insurance line earned 20 million of premium in its first year. With essentially no credible development history, the actuary adopts an expected loss ratio of 70 percent from the pricing analysis, setting ultimate losses at 14 million. With only 0.5 million paid so far, the reserve is 13.5 million — driven entirely by expectation, not by the thin actual data.
Ultimate = earned premium x expected loss ratio; the method ignores actual emerging claims entirely.
Because it ignores actual experience, the ELR method is stable but unresponsive: if the year is genuinely worse than the assumed ratio, the method will not warn you. It is a starting point for green years, not a final answer for mature ones.