IBNR Reserve Calculator for Self-Funded Health Plans

What is IBNR and how is it calculated?

IBNR is the estimate of claims your plan incurred on or before the reporting date but has not yet paid. Care happened; the bill hasn't arrived, or it arrived and hasn't cleared. Under accrual accounting it's a liability whether or not anyone has invoiced you.

The completion factor method works backwards from payment patterns. For any incurred month, some fraction of the eventual total will have been paid by now — that fraction is the completion factor. Divide paid-to-date by the completion factor and you get ultimate incurred claims. Subtract what you've paid and the remainder is IBNR.

A published example makes the mechanic concrete: an October incurred month that paid $40,000 in the first month against an eventual $70,000 total has a lag-zero completion factor of 57%. Apply that factor to a fresh October-like month showing $40,000 paid and you're back to a $70,000 ultimate.

The calculator's defaults run six incurred months. The oldest is 99.50% complete with $1,180,000 paid, giving an ultimate of $1,185,930 and just $5,930 of IBNR. The newest is 40.20% complete with $610,000 paid, giving an ultimate of $1,517,413 and $907,413 of IBNR. Total reserve before margin: $1,569,124.

Note what IBNR is not. It isn't claims you've received and not yet paid, it isn't the stop-loss receivable, and it isn't a rainy-day fund. If your finance team blurs those together the balance sheet conversation gets messy quickly.

Where does most of the uncertainty in an IBNR estimate come from?

The newest incurred months, and it isn't close. Those months have the least run-out data, carry the most actuarial judgement, and are where manual overrides to the mechanical result usually get applied.

The calculator quantifies it directly: 84.12% of the reserve sits in the two newest months. That single figure should determine where your review time goes.

Test it by moving completion factors one at a time. Shift the newest month's factor from 40.20% to 43.00% — under three percentage points — and the reserve drops by $103,749. Move it the other way to 38.00% and it climbs $92,243. A six-point band around your best estimate spans roughly $232,000 of booked liability.

Now do the same to the oldest month. Moving it from 99.50% all the way to 100.00% changes the reserve by $6,226. Dropping it to 98.00% moves it $19,060. The mature end of the triangle barely matters.

That asymmetry is the practical takeaway. Reviewers instinctively check the numbers they understand best, which are the settled old months. The money is at the other end. When you challenge an actuary's estimate, challenge the bottom-right corner of the triangle — the factors for the most recent two or three incurred months — and ask what data supports them.

How do I sanity-check an IBNR estimate before booking it?

Two independent cross-checks, both of which the calculator runs automatically. A completion factor result you haven't corroborated is a number, not an estimate.

The PMPM check. Divide total ultimate incurred by total member months and compare against your own trend assumption. At the defaults, $7,924,124 over 14,800 member months is $535.41 PMPM against an expected $520.00 — a 2.96% divergence, inside tolerance. Widen the gap and the check fires: an expected PMPM of $480 produces an 11.54% divergence and a flag reading "Investigate the delta." This is standard practice in a good actuarial report, and the delta should be explained rather than ignored. If the two numbers diverge materially and nobody acknowledges it, the estimate isn't ready to book.

The loss ratio check. Multiply total plan funding by an expected loss ratio — typically 80% to 90% for a self-insured plan, usually supplied by your broker — and compare the implied IBNR. At the defaults, $9,200,000 of funding at 86% gives expected claims of $7,912,000 and an IBNR of $1,557,000 against the completion factor answer of $1,569,124. A $12,124 difference on a $1.5 million reserve is corroboration.

Change the funding assumption to $8,500,000 and the loss ratio method drops to $955,000 — a $614,124 gap that trips the flag. When two methods disagree by that much, one of your assumptions is wrong and you need to find out which before the auditor does.

What is a margin for adverse deviation and how large should it be?

A deliberate cushion added on top of the mechanical estimate, covering experience the model didn't anticipate — a claims processing backlog, a large case that hasn't surfaced, a shift in provider billing behaviour.

The calculator's default is 5%, taking the reserve from $1,569,124 to $1,647,580. Sweep it and you see the shape:

• 3% gives $1,616,198
• 10% gives $1,726,036
• 15% gives $1,804,492

There's no standard percentage, and anyone who tells you there is should be asked for a source. The size depends on the volatility of your block, how much run-out history supports the factors, and whether anything unusual happened in claims processing during the period. A plan with 36 to 48 months of stable history warrants a smaller margin than one with 14 months and a TPA conversion halfway through.

Two benchmarks help calibrate. The calculator reports the reserve in months of claims — 1.25 at the defaults — and as a percentage of annualized claims, 10.40%. Both are easier to defend to an auditor than a raw dollar figure, and both are easier to compare year over year.

The honest framing is that the margin is a judgement about uncertainty, not a fudge factor for a number you don't like. Set it from the characteristics of your data, document why, and apply it consistently. A margin that moves every quarter to smooth reported results is an audit finding waiting to happen.

Can I do this without an actuary?

You can run the monthly refresh. You cannot build the triangle, and the calculator is deliberately scoped to that split.

Constructing completion factors requires a lag triangle — paid claims arrayed by incurred month against months of run-out, typically 36 to 48 months of history so the pattern stabilises — and then a decision about how to average the observed factors. That averaging choice is genuinely consequential and there are several defensible approaches; it's the part of the job that carries professional judgement and, where a formal opinion is required, professional standards.

What happens monthly is different. The point of a quarterly refresh is not to rebuild the triangle from scratch; it's to reapply the existing completion factors to a fresh paid-claims diagonal and flag any drift. That's this calculator. Take the factors from your actuary's most recent study, enter the current paid figures and member months, and you have a defensible interim estimate between studies.

Two practical notes. Ask your actuary for the triangle itself — it's the evidentiary basis for the estimate, and reluctance to share it is a finding in its own right. And watch the data quality flag: leave a completion factor at zero and that month drops out of the calculation entirely, which at the defaults would understate the reserve by more than $950,000. The calculator catches it and says so rather than quietly returning a smaller number.

If your plan files a formal Statement of Actuarial Opinion, that's a different exercise governed by its own standard. Ask whether one is in scope.