A claim that circulates whenever actuarial hiring comes up is that a wave of career changers displaced by the tech downturn has refilled the entry-level pipeline and made the first job harder to get. That is a claim about supply, and supply is measurable. The dataset behind this site's exam pass-rate explorer records the candidate count at every sitting of 74 CAS and SOA exams, some of them back to 2007. The wave is there. It has also not moved as far down the pipeline as the argument assumes.
The Front Door Refilled, and the Next Room Did Not
Exam P drew 8,735 candidates across its six 2022 sittings, its lowest annual total in over a decade. In 2025 it drew 13,182 across the same six sittings, an increase of 50.9%. Exam FM went from 7,590 to 11,305 over the same window, up 48.9%. Neither exam changed its number of sittings between those years, so the comparison is like for like: half again as many people sat the entry pair in 2025 as in 2022.
The size of that jump is the part worth holding onto, because the destination is unremarkable. Exam P's 2025 volume of 13,182 is 94.5% of its 2018 level and 99.8% of its 2019 level. FM's 11,305 is a fraction above its 2019 figure of 11,267. Both exams ran considerably hotter in the mid-2010s, when Exam P cleared 18,000 candidates in 2013 and stayed above 15,000 through 2017. The correct description of 2022 to 2025 is a recovery from an unusually thin period, not an unprecedented flood. The profession has seen this many people at its entry exams before, and recently.
Exam P and Exam FM are the shared entry pair. A candidate heading for Fellowship in either society sits both before anything else, which means their volume measures general interest in the profession, and it also captures students who sit a prelim and never pursue the credential. The first exam that requires a candidate to have chosen a track is where commitment actually registers, and that is where the growth thins out.
MAS-I, the first CAS-only exam, went from 1,661 candidates in 2022 to 2,145 in 2025. That reads as a 29.1% increase until you account for the CAS running three MAS-I sittings in 2025 against two in 2022. Per sitting, MAS-I fell 13.9%. MAS-II, which also gained a third sitting, fell 5.7% per sitting. Exam 5 held at two sittings in both years and rose 21.3%, but Exam 5 is an upper-level ACAS exam whose candidates entered the pipeline years earlier.
| Exam | 2022 candidates | 2025 candidates | Sittings | Annual change | Per-sitting change |
|---|---|---|---|---|---|
| Exam P (shared entry) | 8,735 | 13,182 | 6 → 6 | +50.9% | +50.9% |
| Exam FM (shared entry) | 7,590 | 11,305 | 6 → 6 | +48.9% | +48.9% |
| CAS MAS-I | 1,661 | 2,145 | 2 → 3 | +29.1% | −13.9% |
| CAS MAS-II | 1,166 | 1,649 | 2 → 3 | +41.4% | −5.7% |
| CAS Exam 5 (upper level) | 1,440 | 1,747 | 2 → 2 | +21.3% | +21.3% |
Read as a funnel, the gateway grew by half while the first track-specific exam grew by less than a third on an annual basis and shrank at every individual sitting. Whatever is arriving at Exam P is not yet arriving at MAS-I in the same proportion.
The Volume Did Not Move the Pass Rate
The intuitive next step is to expect a crowd of underprepared newcomers to drag pass rates down. Exam P says otherwise, in a way that is easy to miss. In 2019, 13,213 candidates sat Exam P and 45.0% of them passed. In 2025, 13,182 candidates sat it and 45.0% passed. The two years differ by 31 candidates and by two hundredths of a percentage point. These are aggregates computed across all six sittings in each year rather than figures the SOA publishes as annual totals, but the arithmetic is simply total passes over total candidates.
What that comparison exposes is that the thin years were the anomaly, not the present. Exam P's raw pass rate reached 51.8% in 2020, the highest annual figure anywhere in its 2007 to 2025 history, on one of its smallest modern cohorts. FM behaved similarly, at 53.8% in 2020 against 48.6% in 2018. A reduced cohort passing through a disrupted testing period cleared these exams at rates neither had produced in normal conditions. The subsequent volume recovery did not degrade anything; it returned both exams to roughly where they sat in 2019.
That matters for anyone using pass rates as an input rather than reading them as news. A hiring plan that assumes some number of analysts will clear a prelim this year, a university program's advising, or a study-time estimate calibrated on 2020 or 2021 figures is calibrated on the loosest conditions in the record and will over-predict how many candidates get through. On Exam P the correction is roughly seven points of raw pass rate between 2020 and 2025. FM is the one exam where the crowding argument finds any support, and even there it is modest: at essentially the same volume as 2019 it passed 44.7% against 48.5%, a decline of 3.8 points. That is real, and it is far smaller than the swings the CAS upper-level exams routinely produce under consistent-standard grading.
What a Candidate Count Is Not
Three limits sit on top of these numbers, and all three cut against reading them as a headcount of new entrants.
A sitting is not a person. Candidates who fail and re-sit appear again, so a rising count is partly the same people cycling through. A 51% rise in sittings is an upper bound on the rise in distinct candidates, not an estimate of it.
The funnel reading may just be a lag. Someone who sat Exam P for the first time in 2023 could reasonably reach MAS-I in 2025 or 2026, so part of the gateway surge may still be in transit. The CAS calendar change cuts the same way: spreading a given annual population across three sittings instead of two lowers the count at each one by construction. The defensible claim is that CAS entry volume has grown more slowly than gateway volume so far, not that the crowd has permanently stalled.
2026 is incomplete and does not reverse the picture. The dataset carries two 2026 sittings for each prelim, so annual totals are not comparable. Set against the same two sittings a year earlier, Exam P is up 6.5% (4,480 against 4,207) and FM is flat at plus 0.7% (3,440 against 3,417). The surge has stopped accelerating and has not begun to reverse.
None of this speaks to the demand side, which is where the original complaint actually lives. Exam volume measures how many people are trying, not how many are being hired, and the two can move in opposite directions for years. What the data settles is the supply half, and it settles it more precisely than the anecdote does. The crowd at the entry exams is real and it grew fast, but it is about the size of the 2019 crowd rather than an unprecedented one, it has not yet shown up in the same proportion at the first exam that requires choosing a track, and it has not made the prelims measurably harder to pass than they were before the pipeline thinned.