Setting Reserves When Severity Keeps Rising

A reserve set at first report is a forecast. It predicts what a claim will cost when it resolves, often years later. In most lines, a careful forecast built on recent comparables holds up. In medical malpractice right now, it tends to run low, and the error points the same way case after case.

This post is for the people who set and defend those numbers: adjusters, claims managers, and defense counsel. It picks up where our post on rising malpractice payouts left off, with the practical question. If severity keeps rising, how should you reserve?

What the latest numbers say

The most recent public read on the line comes from S&P Global Market Intelligence, as reported by Insurance Journal in April 2026. Four points matter for reserving:

• Medical professional liability is the most severity-pressured line in casualty insurance. Occurrence severity reached $151,768 per open claim in accident year 2018, the highest figure across all casualty lines.
• The calm in recent years is not real. The report says the "moderation in newer years" mainly reflects claim immaturity. Recent accident years have not finished developing.
• Large payments carry more of the total. Using federal malpractice payment data, the report found that payments of $500,000 or more made up 36.5% of total malpractice payments in 2024, in 2025 dollars. That is a new high for the line.
• The line is losing money. S&P estimates a calendar-year combined ratio above 105% in 2025, the fifth time in eight years.

Put together, these say one thing. A reserve built on recent averages starts low and stays low.

1. Put a date on every benchmark

Every reserve rests on a benchmark, and every benchmark comes from a window of time. Ask which accident years produced it and whether those years have matured.

A benchmark built on 2022 and 2023 claims is built on the claims that resolved fastest. Those tend to be the smaller ones. The large, slow claims from those same years have not paid yet. When they do, the average for those years will rise.

Treat any benchmark without a date as unfinished. Treat one drawn mostly from recent accident years as a floor, not a midpoint.

2. Reserve to the year the claim will settle

A claim reported today will not resolve today. If severity keeps climbing while the claim is open, the payment will reflect the year it settles, not the year it was reported.

Ask how long claims like this one usually take to resolve. Then reserve to that year. The Hooper Engine, the model behind MedMalPredict, projects its estimates forward to a future settlement year for this reason. A MedMalPredict report also shows the typical resolution timeline for comparable cases, so you can see how far forward you are forecasting.

3. Reserve to the venue, not the national policy

Where a claim is filed moves its value more than almost anything else. In the historical data, the median payout in California is $73,000. In Pennsylvania it is $507,000. Those are comparable cases, about seven times apart.

The S&P report shows the same pattern in carrier results. In 2025, the direct incurred loss and defense cost ratio was 128.8% in New Mexico, 143.8% in Utah, and 125.7% in South Carolina, against a U.S. aggregate of 75.9%.

A single reserving standard applied across states is wrong in opposite directions at once. Start from the venue. Then adjust for the facts.

4. Reserve the tail, not just the middle

Two claims can share the same expected value and carry very different risk. One clusters tightly around the median. The other has a long right tail, where a real share of similar claims paid far more.

The 36.5% figure above is a warning about the tail. When large payments make up a growing share of the dollars, a reserve set to the middle of the range misses where the money goes.

Ask for the spread, not just the midpoint. As we argued in Why "What's My Case Worth?" Is the Wrong Question, a single number hides the shape of the outcome. A payout range and an outcome severity distribution tell you whether you are reserving a narrow claim or a wide one. For a wide one, ask a direct question: what share of similar claims paid more than this reserve?

5. Price the known multipliers at first report

Some facts move value in ways the historical data has already measured. Put them in the reserve on day one, not after the first mediation.

• Severe survival, not death, is the bigger exposure. Cases ending in grave permanent injury have a median payout of about $807,000. Cases ending in death have a median of about $311,000. A reserving habit that treats death as the ceiling is upside down.
• Prior history is a number. Practitioners with five or more prior adverse reports settle for about 43% more than first-time reporters. If the file shows that history, price it.

These are not soft factors. They are measured patterns, and the other side may already have them in its number.

6. Re-test the reserve at each milestone

A reserve is a hypothesis with a date on it. Test it again when the facts change: after key depositions, after expert disclosures, and before mediation. Each time, ask the same three questions. Has the venue read changed? Has the severity read changed? Has the claim moved toward the tail?

A reserve that has not been re-tested since first report is not conservative. It is old.

What a prediction can and cannot do here

A MedMalPredict prediction gives you three things that bear directly on a reserve: the payment probability, a payout range in current dollars, and the distribution of outcome severity across comparable cases. It draws on 270,000+ resolved malpractice claims from 2004 to 2026, across all 50 states and U.S. territories, and returns results in seconds.

It does not replace your actuaries. It speaks to the payment on a single claim, not to defense costs, incurred-but-not-reported losses, or the aggregate picture of a book. It is a statistical estimate, not legal advice. Your judgment about the facts, the witnesses, and the people still decides the number.

What it can do is check the anchor. Before a reserve goes into the file, you can see whether it sits in the middle of what similar claims have paid, or well below it.

The short version

Severity is rising, and the recent data understates how much. Reserve to the settlement year, to the venue, and to the tail. Price the multipliers you already know about. Then test the number again whenever the case changes. The goal is not a high reserve or a low one. It is a reserve that holds up when the claim finally resolves.