Unum Group (NYSE: UNM) — The Easy Half of the Long-Term Care Block Has Been Sold

Report date: 2026-07-18 · Price: $89.05 (2026-07-17 close) · Market cap: ~$14.3bn · Shares out: 160.7M Sector: Financials · Insurance — Life & Health (Group Disability / Life / Supplemental) Coverage status: Initiation

Sections 1–15 of this article carry no investment recommendation and no price target. The single, deliberate exception is the Claude's Take block immediately below, which is clearly labeled as such.

⚡ Claude’s Take

The author’s own independent opinion, offered as general information and not investment advice. The analytical body (Sections 1–15) below carries no position, no recommendation, and no price target.

Verdict: HOLD / AVOID at $89 — accumulate on weakness toward ~$72–78 (roughly 0.92–1.00x ex-AOCI book, ~10.0–10.9x normalized EPS). Conviction: medium.

Tag: “They sold the half that had a bid.”

Unum spent a decade trading at 0.4–0.9x book because the market treated its long-term-care block as unquantifiable. Two Fortitude Re cessions later — ~40% of LTC statutory reserves gone in eighteen months — the market now treats that block as quantified and being transferred, and has re-rated the stock from 0.67x ex-AOCI book (FY2022–23) to 1.13x today, an all-time-high $92.73 in June and the 98.4th percentile of its own ten-year valuation range. That re-rating was legitimate. My problem is that it is essentially finished, and the evidence says the next leg requires something that has never happened. After the July 2026 deal closes, ~$11.0bn of LTC statutory reserves remain and ~70% of it is GROUP long-term care. Every completed LTC block transaction in the industry since 2023 — MassMutual/RGA, Lincoln/Fortitude, Manulife/Global Atlantic, Manulife/RGA, Brighthouse/Aquarian, and both of Unum’s own — has been individual or hybrid. There is no observed bid for group LTC anywhere in the market. Unum sold the half that had a buyer.

Underneath, three things are moving the wrong way at once and none of them is in the headline number. The LTC net premium ratio is 97.6%, up from 93.5% in 2023, with ~2.4 points of headroom before the 100% cap converts the block into a one-way instrument where adverse deviation hits earnings immediately and favorable deviation cannot be banked — Q1-26’s −$145.3M Closed Block operating loss was the first observation of that regime, and the stock made an all-time high ten weeks later. The core benefit ratios have already turned (group disability 59.0% in FY24 → 62.4% → 63.7% in Q1-26, with management itself guiding to ~65% “over the next few cycles”), and the Q1-26 “beat” was carried by a group-life ratio of 61.8% against a 68–72% guide that the CFO called “really an anomaly.” And the capital return — ~$1.0bn buyback plus ~$300M dividends — is running roughly $500M a year ahead of 2026 ordinary subsidiary dividend capacity of $631M, bridged by regulator-approved extraordinary dividends and a deliberate draw-down of RBC from 440% to 400–425%. Management then redefined adjusted operating income in Q1-2026 to exclude the entire Closed Block, in the exact quarter that block lost $145.3M. On the metric the market is capitalizing — management’s $8.13 — that is not what this company earns. Two independent workstreams put defensible normalized EPS at ~$7.00–7.20, an ROE of ~9.1% against a cost of equity of ~9.5%. A business earning its cost of capital is worth about book.

The honest counterweight, which I will not bury: on ex-AOCI book UNM is the cheapest name in its own factor cohort (1.13x vs PRU 1.17, MET 1.30, CNO 1.35, VOYA 1.38, RGA 1.44), it screens ~8% below a cohort ROE/P-B regression line, and the sum-of-the-parts shows the market marking the Closed Block at roughly zero — conservative, not euphoric. This is not a bubble; it is a completed repair trade priced at the top of its base case. The framing is neither momentum nor falling knife: FactorsToday shows a zero momentum loading despite the all-time high, with DividendYield (+0.42) and Value (+0.21) the dominant style exposures — both at ~+1.7 standard deviations of favor. It is a finished value re-rating now held as an income/defensive carry trade, in a factor regime that flatters exactly that profile. I want to own this business; I do not want to buy the last 20% of a re-rating whose next chapter requires a market that does not exist.

What flips me bullish: a signed group-LTC risk transfer at a premium no worse than the ~27% Unum paid on Fortitude II — that single event breaks the whole bear case and I would pay up immediately. What flips me bearish: the net premium ratio reaching the 100% cap (on current drift, ~18 months), or a Q3-2026 assumption review that strengthens LTC reserves for the third time in four years while the group block remains untradeable.

📈 Stock Price Action — Five-Year Event Map

Unum has round-tripped from an ~$8.25 COVID-crash low (2020-03-23, a −65% drawdown from the pre-COVID $23.60) to an all-time high of $92.73 on 2026-06-12 — the highest split- and dividend-adjusted close in the price history’s entire 1986–2026 record. The stock closed at $89.05 on 2026-07-17, 3.97% below that all-time high, inside a 52-week range of $67.06–$93.22, and above its 21-day ($88.95), 50-day ($86.52) and 200-day ($79.46) moving averages. Trailing returns: +12.5% (3m), +18.4% (6m), +12.6% (12m), +98.8% (3y), +284.9% (5y) — the last leg delivered while FY2025 GAAP EPS fell 55% to $4.27 and GAAP ROE fell to 5.6%.

1. The February–March 2020 collapse took the stock to $8.25, a −65% drawdown in five weeks — deeper than the S&P’s, consistent with a carrier whose legacy LTC block makes it a levered proxy for credit and discount rates (INTERPRETATION). 2. The recovery was among the largest in the S&P 500 over that span: +221% off the low, with a +14.3% single session on 2020-11-09 (vaccine efficacy day) — a factor-rotation move, not a company event (INTERPRETATION). 3. The 2021 give-back (−26% to a $19.57 low on 2021-12-01) coincided with variant waves, elevated mortality claims and a renewed rally in long rates; the LTC-overhang discount reasserted itself (INTERPRETATION). 4. The 2022–2023 re-rating (+134%) tracked the rate-hike cycle and a restarted repurchase program (INTERPRETATION). The day-level catalysts for the two +14% sessions in May and August 2022 are not confirmed from primary sources and are treated at period level. 5. The 2024 run to $74.28 was earnings-led — FY2024 GAAP EPS of $9.46 on ROE of 14.6% — though Section 6 below shows both figures were materially flattered by a +$357.4M pre-tax reserve assumption benefit that reversed the following year (price: FACT; the flattering decomposition: FACT; the market’s reaction to it: INTERPRETATION). 6. The single worst session of the five years was 2025-07-30, −12.2%, the day after Unum reported Q2-25 adjusted EPS of $2.07 against $2.23 consensus. The drawdown extended to $67.36 on 2025-08-18, −15.7% from the February 2025 high. 7. The decisive price-action fact of this article. On 2025-11-03 Unum reported Q3-25 net income of $39.7M versus $645.7M a year earlier — a 94% decline driven by a $643.1M pre-tax Closed Block LTC charge. The stock closed up 7.2% the next session and has never traded back to the pre-print level; it now sits ~23.6% above it. On the call, analyst Michael Burdis (Raymond James) offered — and CFO Steven Zabel accepted verbatim — the framing that the charge carried no cash impact, implied more future cash, and reduced the likelihood of future charges. That exchange is the cleanest available explanation for why a 94% earnings collapse produced a positive move (INTERPRETATION). 8. The 2026 leg (+30% from 2026-02-27 to the 2026-06-12 all-time high) followed a ~10% dividend increase, a Q1-26 net income improvement to $232.0M, and the reclassification of Closed Block results below the adjusted-operating line. The 8-K filed 2026-07-06 accompanied the $3.8bn statutory-reserve LTC cession to Fortitude Re, on which the stock fell 3.1% (FACT) — the first datapoint suggesting the market may be starting to scrutinize the price of de-risking rather than applaud the fact of it (INTERPRETATION).

1. Executive Summary

Unum Group is the largest or second-largest writer of US group long-term disability, a top-10 group life writer, and the owner of Colonial Life, a genuinely differentiated worksite voluntary-benefits franchise. Its three core operating segments earned $1,489.5M of after-tax adjusted operating income on $7,251.2M of allocated equity in FY2025 — a 20.5% ROE. That is a good business. The problem is that the entity a shareholder actually buys also contains a run-off long-term-care block that consumed $4,830.0M of equity — 38% of the group total — to earn $40.7M, a 0.8% return. Consolidated, Unum is an 11.1% ROE company, and on a properly normalized basis a ~9.1% ROE company against a cost of equity of roughly 9.5%.

FY2025 made the tension explicit. GAAP net income fell 58.5% to $738.5M ($4.27 diluted, from $9.46), and GAAP ROE fell to 5.6% from 14.6%. The cause was a $478.5M pre-tax / $377.8M after-tax net reserve assumption charge, of which Closed Block LTC alone was −$643.1M while every core segment posted a benefit. This was the third LTC charge in four annual reviews (FY2023 −$368.1M, FY2024 +$174.1M, FY2025 −$643.1M; cumulative −$837.1M over three years). Reserve assumption updates appear in five of the last five years, and they do not average to zero: the pre-tax mean is −$99.4M per year. Management excludes them from its headline metric, and from executive compensation.

Management has been genuinely, materially active on the problem. Two coinsurance transactions with Fortitude Re — $3.4bn of statutory reserves ceded in 2025, a further $3.8bn announced 2026-07-02 — have removed roughly 40% of LTC statutory reserves in eighteen months. The skeptics who said the block-transfer market was closed to Unum were wrong. But the price did not improve on the second bite (~17% premium over ceded GAAP reserves on Fortitude I; ~27% over best-estimate reserves on Fortitude II, with ~$5.7bn of assets transferred against ~$4.5bn of best-estimate liability), the cost of reinsurance is now a permanent $116.7M pre-tax / $0.53 per share annual drag, and — decisively — both deals were individual LTC. The ~$11.0bn residual is ~70% group LTC, a category with no observed buyer in any completed industry transaction.

Meanwhile the core is past its cycle peak. Group disability’s benefit ratio has risen from 59.0% (FY2024) to 62.4% (FY2025) to 63.7% (Q1-26), with management guiding to 62–64% near-term and ~65% “over the next few cycles.” The UK book deteriorated 580bp. Group life at 61.8% in Q1-26 is running six points better than management’s own 68–72% guide on what the CFO called “really an anomaly.” Persistency fell across every US group line in FY2025 (LTD 93.3%→91.1%, STD 91.7%→88.9%, group life 92.0%→90.2%) as Unum pushed rate, and LTD sales fell 17.7%. Over 2019–2023 Unum’s LTD in-force grew at a 1.35% CAGR against a market growing 3.42% — the second-slowest of eleven carriers.

At $89.05 the stock trades at 1.31x GAAP book but 1.13x book excluding a −$1,795.5M AOCI position — and at the 98.4th percentile of its own ten-year valuation range. A sum-of-the-parts shows that at any core multiple above ~11.4x, the market is ascribing zero or negative value to the Closed Block: conservative, not euphoric. But the reconciliation of that observation with the record percentile is the intellectual core of this article. The market has repriced the LTC block from unquantifiable to quantified and partially transferred. That was correct and it is largely done. The remaining upside requires the group block to trade.

No recommendation and no price target appear anywhere in Section 1–Section 15.

2. Business Overview

2.1 What Unum does

Unum Group, founded 1848 and headquartered in Chattanooga, Tennessee, is a workplace-benefits insurer with roughly 10,910 employees. It sells financial-protection products to employers, for the benefit of their employees, through in-house field sales teams, independent brokers and consultants, and — at Colonial Life — an independent-contractor career agency force. It reports four segments plus Corporate.

Unum US (~72% of core operating income) writes group long-term and short-term disability, group life and AD&D, and a supplemental & voluntary book comprising individual disability (multi-life), voluntary benefits (accident, critical illness, hospital indemnity), and dental & vision (through the Starmount platform acquired in 2016).

Colonial Life (~26%) sells voluntary/worksite accident, sickness, disability, life and cancer/critical-illness products, employee-paid, at the worksite, through its own captive agency force — over 87,000 business clients and 4M+ covered workers.

Unum International (~9%) is principally a UK group risk business (group income protection, group life, critical illness) plus a smaller Poland operation.

Closed Block is the run-off: legacy individual long-term care, group long-term care, and the legacy individual disability block. It writes essentially no new business and, effective 2026-02-01, no longer accepts new enrollments even on existing group LTC cases — a deliberate tail-risk reduction that management describes as eliminating tail risk and that a Bank of America analyst characterized on the Q3-25 call as forgoing positive cash flow “because you simply want to make the book smaller.” Management confirmed: “That is correct.”

2.2 Segment economics

Adjusted operating income by segment (pre-tax, $M) — FACT, FY2025 10-K and statistical supplements:

Consolidated P&L, insurance format ($M) — FACT, FY2025 10-K p.105:

Note the remeasurement line: −$54.8M, −$562.3M, +$439.1M. That single line — the annual reserve assumption review plus current-period experience remeasurement — swings by a billion dollars across three years on a company that earns roughly $1.4bn. It is the most important number in Unum’s income statement and management reports it below its headline metric.

2.3 How the money is actually made

Decomposing FY2025 pre-tax adjusted operating income by source is unflattering to the “underwriting franchise” story:

Consolidated net investment income of $2,032.7M is 114% of consolidated adjusted operating income, and roughly half of all NII sits in the run-off Closed Block ($1,016.5M), where it is declining (−15.1% year-over-year in Q1-26). Under LDTI a portion of that NII is contractually owed to policyholders through the reserve discount and is therefore already inside Policy Benefits, so the 47.5% figure is an upper bound on the rate-sensitive share of Unum US profit — but directionally the point holds and matters: roughly half of Unum US’s earnings are spread income on float, which carries no competitive advantage and reverses when rates fall. Colonial Life, at 62.4% underwriting-driven, is the higher-quality earnings stream in the group.

2.4 Persistency — the recurring-revenue question

Group benefits are yearly renewable term. The entire book is re-priced annually and can be re-bid annually. Persistency is therefore the single best measure of whether this is a recurring-revenue business or an annual auction.

Verdict: Unum is a scaled, diversified, genuinely profitable workplace-benefits underwriter whose reported earnings are roughly half float income, whose group book turns over about one case in ten every year, and whose consolidated returns are dominated by a run-off liability that is 38% of the equity. The operating business is sound. The entity is a hybrid of a good insurer and a large, slowly-resolving actuarial problem, and any analysis that looks only at the former is looking at 62% of the balance sheet.

3. Industry Dynamics

3.1 Market structure and size

US group disability (LTD + STD + PFML) in-force premium was ~$19.9bn in 2024 on the Milliman survey panel, growing +4.7%; LTD in-force +4.2%, STD +5.2%, PFML +3.3%. New sales premium was $2.7bn (+8%). Group term life in-force premium was ~$32.8bn (Gen Re 2025 survey, +1% year-over-year), with new sales of ~$3.0bn and top-10 carriers at 84% of new sales. On the NAIC’s broader statutory basis, industry group life direct written premium was $44.7bn in 2024.

Group LTD in-force league table (Milliman, 2024 data): Unum $2,004.6M, Lincoln $1.9bn, MetLife $1.9bn. Unum ranks #1 on all four Milliman cuts (LTD in-force, STD in-force, LTD new sales, STD new sales).

A critical methodological catch that must not be glossed: the 2025 Milliman survey (2024 data) dropped The Hartford, Sun Life, and Reliance Matrix from the panel. In the 2024 survey (2023 data), Hartford ranked #1 in LTD in-force at $2,179.0M, ahead of Unum’s $1,946.1M. Unum’s apparent 2024 #1 ranking is therefore substantially a disclosure artifact, not a share gain. Hartford’s own FY2025 disclosures show Employee Benefits premium flat-to-growing, so Hartford very likely remained #1 or co-#1. The honest statement is that Unum is #1 or #2 in US group LTD and has been for six years, too close to call.

CR10 in workplace life new sales is ~64%; top-10 disability carriers hold ~75% of new premium. This is a moderately concentrated oligopoly, not a fragmented market — and not an oligopoly with pricing power, for reasons developed below.

3.2 The pricing cycle — the crux, and it has already turned

Three independent tailwinds carried group-benefits margins from 2021 to 2024. All three peaked together in FY2024 and are now reversing.

(a) Group life mortality. Post-COVID working-age mortality normalized and then over-shot favorable. Unum’s group life & AD&D benefit ratio: 78.7% (FY22) → 72.7% (FY23) → 66.3% (FY24) → 67.5% (FY25) → 61.8% (Q1-26). Peers confirm: Hartford’s group life loss ratio fell 83.5% → 78.7% → 76.3% → 73.2% (Q1-26); MetLife’s group life mortality ratio printed 80.1% in Q1-26 against a 2026 target range of 83–88% and it cut the range by a point on the view the trend persists; Lincoln’s life loss ratio fell 74.0% → 66.9%. This is a shared, free, non-differentiating benefit. Critically, the Hartford’s FY2025 10-K states the favorable group life mortality “is not expected to repeat in 2026” — the opposite read to MetLife’s on the identical variable. Unum’s own CFO called Q1-26’s 61.8% “really an anomaly” against a 68–72% guide, noting the historical average is “in that high 60% range.”

(b) Disability incidence and recovery. This has already mean-reverted, and management says so.

Unum’s stated drivers in Q1-26: “higher incidence in the short-term disability product line” and, in Q4-25, “lower claim resolutions driven by mortality in group LTD.” Hartford names “higher short-term disability claim incidence, including PFML.” CFO Zabel has guided the ratio to “glide up to that 65% range” over the next few cycles. This is management telling the market the trough is behind it.

The mean-reversion risk is under-appreciated and structural. Disability claiming is counter-cyclical — in the Great Recession, SSDI applications rose by roughly one million with 41.8% awarded; RGA’s published work on private group states plainly that “claim costs tend to increase during recessions — new claim submissions rise while claim recoveries fall for group LTD insurers.” Unemployment is ~4.9% and LIMRA projects employment growth below 1% for 2026–2028. And STD leads LTD by the elimination period (90–180 days) by construction — so Unum’s own Q1-26 disclosure of rising STD incidence mechanically foreshadows rising LTD incidence into late 2026 and 2027. Management’s 62–64% guide has never been tested against a genuine unemployment cycle.

© Discount rates. New LTD claim reserves are struck at current single-A yields under LDTI; higher rates mechanically lower the reserve and flatter the reported benefit ratio. Unum’s disclosed “Effect of Change in Discount Rate Assumptions on the LFPB” fell from $1,185.4M ($6.64/share) at 12/31/24 to $929.9M ($5.61/share) at 12/31/25 — roughly $255M of rate-driven reserve relief handed back in a single year. NII is falling across every line (group disability −5.5%, group life −2.8%, supplemental −5.2% in Q4-25).

3.3 Is capacity leaving or arriving?

Neither, in the traditional lines — and that is the problem. There is no evidence of any carrier exiting US group life or group disability in 2023–2026. What is happening is the opposite: incumbents are redeploying into the pool. MetLife’s “New Frontier” strategy tilts explicitly toward Group Benefits; Securian’s record 2025 was “anchored in group benefits”; Sun Life claims #1 US group benefits and grew US group sales +27% to a record $1.8bn; Mutual of Omaha grew group premium +43% in two years at a 17% five-year CAGR; StanCorp/The Standard has bought its way to scale three times in four years (Securian group pension 2022, Elevance group life 2024, Allstate’s Employer Voluntary Benefits for $2.0bn, closed April 2025). Tokio Marine’s newly-disclosed IFRS “Employee Benefits” line runs a 95.3% combined ratio versus 85.4% for its Specialty P&C — confirming group benefits as the lowest-margin block on its North American platform, and it still guides revenue growth for 2026.

The single most telling price datapoint: in 2024, LTD in-force premium-per-life rose +3.5% to $292.3, while new-sales premium-per-life fell −13.5% to $214.2. New business is being written roughly 27% below the in-force rate per life. That is price competition, quantified.

In Marathon’s capital-cycle terms: no new capital, but rising incumbent capacity commitment into the same ~$30–35bn pool at peak reported margin, with the price evidence already showing capacity biting. The absence of new entrants is not the absence of new capacity.

3.4 The one line where capacity IS leaving — medical stop-loss

Adjacent but decisive for reading Unum’s management. The medical stop-loss market ran a record ~91% loss ratio in 2025, having been in the 80s since 2019 (81.6% in 2019, 86.0% in 2024 on Oliver Wyman/Guy Carpenter statutory data). Claim severity has exploded: $1M+ claim frequency +131% since 2022; Tokio Marine HCC reports a 1,251% increase in >$2M claim frequency since 2013. Voya’s 2024 policy-year loss ratio developed from 81% to 94% inside one year, its reported quarterly ratio hit 115% in Q4-2024, its segment CEO departed in the same 8-K that disclosed the warning, and it is now under activist pressure to sell the unit. Sun Life’s ratio went 67% (2023) → 74% (2024) and it declined to disclose FY2025. Gallagher Re reported in February 2026 that three major reinsurers have signaled exit from US medical reinsurance, calling it “a structural — rather than cyclical — adjustment.”

Unum sold its medical stop-loss business to Amynta Group in July 2024, closing Q3-2024 — immediately before the blowup, and while it was still reporting “favorable medical stop-loss benefits experience.” That is a genuinely excellent, well-timed capital-allocation decision and the strongest single piece of evidence for management quality in this file. It is discipline, not moat.

3.5 Long-term care industry

The industry holds >$200bn of statutory LTC reserves against ~5.8M standalone policyholders. Incurred claims of $17bn (2024) are projected to peak at $44bn in 2041; claimant count peaks in 2037. The industry is roughly fifteen years from peak claims — reserve adequacy will not be settled for two more decades.

Carriers are still adding to reserves. Genworth’s Q4-25 review added $286M to LFPB, of which $124M was actual-versus-expected experience variance (“higher claims and lower terminations”) rather than assumption change — the experience miss being the more alarming component, since it means current experience runs worse than already-strengthened assumptions. Manulife/John Hancock took a Q3 LTC charge plus a distribution-intangible write-off.

The rate-increase regime delivers, slowly. Milliman’s filing survey (17 insurers, >75% of industry premium): 73% of filings fully or partially approved, average approved increase 28%, six months to disposition. Requests run far larger than approvals — Unum’s Oregon filing requested 50–150%. The NAIC’s Multistate Actuarial Framework was revised in December 2025 toward a single review methodology, modestly carrier-positive, but participation is voluntary.

The block-transfer market is open and deepening — for individual LTC.

Deal sizes have scaled well past the predicted “bite-sized $1–3bn,” and buyers moved from mature blocks (average attained age 83) to younger ones (age 75) — rising confidence in pricing LTC morbidity. The enabling structure is bifurcation: an asset manager takes the spread risk and retrocedes 100% of the biometric risk to a highly-rated global life reinsurer. Fortitude did exactly this on both Unum deals. The scarce commodity is not capital — it is the handful of reinsurers willing to hold LTC morbidity.

Every single transaction in that table is individual or hybrid. Not one is group LTC. The one counter-signal in the set runs the wrong way: in September 2025, Munich Re recaptured ~$350M of Continental General’s Kanawha LTC reserves — risk flowing back to the cedent.

3.6 Regulation

ERISA preemption is a real but non-differentiating structural subsidy to group disability carriers: remedies are confined to benefits owed plus attorney’s fees, with no bad-faith, punitive or emotional-distress damages, review limited to the administrative record, often under arbitrary-and-capricious deference. In Greenwald’s taxonomy this is government protection — but it protects the whole industry’s profit pool and differentiates no participant. Unum’s 2004–05 multistate market-conduct settlement over claims handling makes enforcement risk a standing item.

LDTI (ASC 944) is why GAAP is close to unusable here without adjustment: annual cash-flow assumption reviews hit net income at the locked-in discount rate, while current-discount-rate remeasurement flows through OCI. This produces both the earnings volatility documented in Section 6 and the AOCI distortion documented in Section 6.3.

UK / Unum International. Swiss Re’s Group Watch shows UK group risk in-force premium grew just 2.0% to £3.67bn in 2025 (from +5.8% in 2024), and group income protection lives actually FELL 1.6% to 3.33M even as policy count rose — the market is shifting to shorter benefit periods for affordability, driven by the April 2025 employer National Insurance increase. Unum grew UK premium +17.1% (Q4-25) and +16.2% (Q1-26) into a 2%-growth market while its UK benefit ratio deteriorated 500–580bp. Growth at eight times market in a shrinking-margin market is usually bought on price or underwriting standards. This is a small segment (~9% of core AOI) but it is a clean warning signal about how Unum is currently competing.

PFML is the rare growth product (~$1.9bn in-force, +3.3%) but benefit design is set by legislatures, there is no credible experience base, and both Unum and Hartford now name PFML utilization as a margin drag. Notably, Unum ranks only #5 in PFML in-force ($205.4M) behind ShelterPoint, MetLife, Lincoln and Guardian, and does not appear among the top PFML new-sales writers — a meaningful failure to convert its claimed leave-management leadership.

3.7 Verdict — structural attractiveness

Mediocre: a structurally average industry at a cyclical peak, carrying a tail liability that periodically confiscates a decade of earnings.

The decisive quantification is Unum’s own segment table: the operating businesses earn 20.5% on $7,251.2M, comfortably above a ~9–10% cost of equity — while the legacy block consumes 38% of group equity to earn 0.8%, halving consolidated ROE to 11.1%. That single table is the industry’s structural signature.

Running Greenwald properly: barriers to entry are partial and weak — real infrastructure, data, licensing and broker-relationship barriers exist (hence CR10 ~64% and zero new entrants), but they are barriers to entering at scale, not to competing for any individual case. Customer captivity is low by construction, since coverage is re-bid annually. The scale-plus-captivity conjunction therefore fails, which is precisely why MetLife, Sun Life, Securian, Hartford, Principal and Mutual of Omaha can all credibly attack the incumbent. Share stability is real (no top-7 carrier moved more than ±1.31 points of LTD share over 2019–2023) but is largely mechanical — yearly-renewable-term products at ~90% persistency make in-force share a slow-moving lagging average, while new-sales rankings churn ±40–60% annually.

Where does value leak? The employer, not the insured, is the buyer — and the employer’s agent, the broker, is compensated on placement and sells the employer on extracting carrier concessions through the annual RFP. The agency relationship that normally creates manufacturer pricing power runs the wrong way here. And broker distribution has consolidated violently into roughly six balance sheets in thirty months (Aon/NFP $13.4bn, MMC/McGriff $7.75bn, AJG/AssuredPartners $13.45bn, Brown & Brown/Accession $9.825bn, OneDigital >$7bn, Hub at a $29bn valuation) — concentrating placement leverage exactly as carriers most need rate.

Cross-industry: group benefits runs 6–10% margins at 15–22% segment ROE. Specialty P&C earns materially more because E&S pricing is not annually re-bid on a standardized RFP and freedom-of-rate-and-form is a genuine regulatory moat. Life/annuity earns less on a heavier, spread-dependent balance sheet the capital cycle is actively attacking. Group benefits sits between: better than annuities on capital intensity, worse than specialty P&C on pricing power and barrier durability.

4. Competitive Position

4.1 Naming the moat — or its absence

Verdict: WEAK / NARROW. There is a real supply-side cost advantage in disability claims management, but it is not paired with customer captivity, and under Greenwald scale without captivity is not a barrier. At the corporate level, Unum does not have a durable competitive advantage.

Supply/cost advantage — PARTIAL, REAL. Unum’s scale in disability claims adjudication, return-to-work management, and disabled-life reserve administration is genuine and shows in margin. After-tax margin on segment revenue: Unum US 12.6% versus Hartford Employee Benefits 8.4% — and Hartford guides its long-run net income margin down to 6–7%. Even Unum’s pure group disability line at a 10.3% margin beats Hartford’s blended segment. This leg passes.

Customer captivity — FAILS, and the evidence is management’s own words. Christopher Pyne, head of Group Benefits, explaining the Q1-26 persistency improvement on the earnings call: “It shows up in terms of persistency, and we can keep customers with a modest rate reduction going forward at very high margins.” The Q1-26 10-Q confirms the group disability benefit ratio worsened to 63.7% “due to pricing actions.” Retention bought with a rate cut is a discount, not captivity. Confirming the elasticity: when Unum pushed rate in FY2025, LTD persistency fell 93.3%→91.1%, STD 91.7%→88.9%, group life 92.0%→90.2%, and LTD sales fell 17.7%. That is a price-elastic book.

Unum’s own 10-K concedes the case. It contains no market-share claim. It lists price among the principal competitive factors. It states plainly: “All areas of the employee benefits markets are highly competitive due to the yearly renewable term nature of the group products and the large number of insurance companies offering products in this market” and — remarkably — “Some of these companies have more competitive pricing.”

Network effects: none. Brand: negligible (the buyer is a benefits manager advised by a broker, not a consumer).

Leave management is Unum’s best moat candidate, and it is converging to table stakes. Management credibly ties it to outcomes: HR Connect customers show “close ratios that are roughly double” and “persistency at levels 2% to 4% higher,” across roughly one-third of core premium; Unum is an inaugural Workday Wellness Partner with API-level HRIS integration. But Hartford has the direct analogue (HartConnect, integrated to bswift, Businessolver and Employee Navigator); MetLife, Lincoln, Sun Life, Reliance Matrix and Guardian all compete; and Sedgwick is a large specialist. Damningly, fee income — the cleanest financial proxy for the leave-management franchise — was $211.6M → $232.1M → $228.6M (−1.5% in FY2025), roughly 3% of Unum US premium and shrinking. A “moat” that cannot be tied to a financial outcome that would deteriorate in its absence is not a moat. And Unum’s failure to convert leave leadership into PFML share (#5 in in-force, absent from new-sales leaders) is direct disconfirming evidence.

4.2 The market-share stability test

Reconstructing a consistent 2019→2023 Milliman panel (all five majors present every year), Unum’s group LTD in-force share fell from 14.75% to 13.61%, −1.14 points. On Greenwald’s literal test (<2 points over 5–8 years = formidable barriers), the industry passes. But the absolute growth rates are the tell:

Unum grew LTD in-force at 1.35% CAGR against a market growing 3.42% — roughly one-third of market growth, and the second-slowest of eleven carriers. Unum’s own FY2025 numbers extend the trend: group LTD premium fell from $2,086.1M to $2,011.1M with LTD sales −17.7%.

This is not a share collapse; it is slow, one-directional erosion inside a stable structure — the profile of an incumbent defending price and margin, not one with a widening moat. The honest nuance: group life premium grew +11.4% cumulatively with sales +18.4%; the medical stop-loss run-off contaminates the disability line; and walking away from underpriced business is correct behavior. But none of that is evidence of a moat. And note who is taking the share: Mutual of Omaha (+10.2% CAGR), The Standard/Meiji Yasuda, Reliance Matrix/Tokio Marine, Guardian — disproportionately mutual and foreign-owned carriers with patient capital and lower return hurdles than a public company targeting 20%+ segment ROE. That is the classic Marathon setup: capital attracted into a business earning above its cost of capital, by competitors who will accept less.

4.3 Colonial Life versus Aflac US

Colonial beats Aflac US on margin and is 62% underwriting-driven — genuinely the better business of the two, and the higher-quality earnings stream within Unum. But it is 26% of core AOI, mid-70s persistency means ~25% of the book must be re-sold annually, and sales grew just +3.8% cumulatively over three years and +0.9% in Q1-26, which the retiring segment president called “a little sluggish” on a “soft pipeline.” Both large captive-agent worksite franchises show the same pattern — the broker/group-sold channel growing, the career-agent channel stagnating, with the enrollment layer intermediated by broker-controlled benefits-administration platforms. The captive agency force is holding the line, not compounding.

4.4 The ROE test — the sharpest finding in this article

Management headlines a 20.5% adjusted operating ROE. That figure excludes 43% of shareholder equity. Unum’s own reconciliation:

Q1-26: core 21.7%, total 7.6% annualized.

The bridge from GAAP 6.7% to the headline 20.5% runs: +6.0 points from excluding $667.5M of after-tax charges, −1.6 points from adding back AOCI, then +9.4 points from dropping the Closed Block’s and Corporate’s capital out of the denominator.

An important correction to a common assumption: for Unum the ex-AOCI adjustment is conservative, not flattering. AOCI is negative $1,795.5M, so excluding it raises the equity denominator and cuts ROE by 1.6 points. The flattery comes entirely from excluding recurring reserve charges and the loss-making block’s capital — not from AOCI.

Unum’s 12-year GAAP ROE average is 9.33% (median 10.66%), and every sub-6% year — 2014, 2018, 2025 — is an LTC charge year. Against a 9–10% cost of equity, Unum has earned approximately its cost of capital across a full cycle and no more.

Management has stated it will exclude the Closed Block from adjusted operating income beginning in 2026 — removing the loss-making segment from the headline metric in the same year its net premium ratio reached 97.5%. Post-2026 series will not be comparable to what came before.

4.5 Verdict

Unum is a well-run, scaled, disciplined underwriter — Greenwald’s “operational effectiveness,” which is emulable — competing in a structurally competitive, low-barrier, slow-growth industry. Roughly half of core-segment profit is rate-sensitive float income carrying no advantage; 38% of shareholder capital is trapped in a run-off block earning 0.8%; the flagship line has lost share for five years; and the one credible moat candidate (leave management) shows shrinking fee income and a #5 rank in the adjacency it should dominate. Unum US and Colonial Life are good businesses. Unum Group, the entity a shareholder buys, is a ~9–11% ROE company with no durable competitive advantage.

5. Growth History and Forward Opportunities

5.1 The historical record

Premium income compounded from $10,046.0M (FY2023) to $10,831.0M (FY2025) — +3.8% annually, essentially the rate of US wage-and-employment growth, which is what a mature group-benefits book should do. Management’s own long-run disclosure is a 10-year core premium CAGR of ~4% and a 10-year ex-AOCI BVPS CAGR of ~8% ($39.24 in 2016 → $78.02 in 2025). The wedge between the two — 4 points a year — is almost entirely the buyback, not organic earnings growth.

Segment detail is more informative than the aggregate:

The flagship is shrinking and everything else is growing. The mix shift out of LTD (long-tail, higher-margin, 20.4% income ratio in FY2023) into STD and PFML (short-tail, lower-margin, faster-growing) is deliberate and it is structurally margin-dilutive: the group disability income ratio fell from 20.4% (FY2023) to 15.2% (FY2025) to 13.2% (Q1-26). That is the mechanism behind the segment’s earnings decline, and it is not a one-off.

Unum US total sales were $1,283.8M → $1,367.0M → $1,358.9M (−0.6% in FY2025). Q1-26 sales jumped +20.8% to $335.1M, concentrated in core-market (<2,000 employees) group sales +31.9% — but Q1 is the smallest sales quarter (Q4-25 alone was $639.9M), so a +20.8% Q1 on a $277.5M base reads through weakly to the year.

5.2 Forward opportunities, assessed honestly

Real: (i) Core/small-case market share — the <2,000-employee segment is where Unum’s service and technology integration genuinely differentiates, and Q1-26 core sales +31.9% is the best evidence in the file that HR Connect is working. (ii) Colonial Life worksite expansion into larger cases and broker-sold distribution, though the +0.9% Q1-26 sales print argues this is aspiration not execution. (iii) Dental & vision, growing 16.8% over two years — but at a 75.6% benefit ratio, the worst line in the company, with FY2025 sales down 11.4%.

Overstated: (i) PFML. It is the industry’s genuine growth product and Unum is not winning it — #5 in in-force, outside the top ranks in new sales, and now naming PFML utilization as a benefit-ratio drag in newly-effective states. (ii) UK international. Growing 16% into a 2%-growth market while the benefit ratio deteriorates 580bp is not growth worth having. (iii) Leave management as a growth engine — fee income is declining.

New and worth flagging: a $500M funding-agreement-backed note program established February 2026 for “spread lending” — levered investment income with no competitive advantage. Nothing drawn. It is the first sign of Unum reaching outside its franchise for earnings. Flag it; do not yet weight it.

5.3 Verdict — high- or low-quality growth?

Low-to-medium quality. The top line compounds at roughly nominal payroll growth, which is what this industry offers and is not a criticism. The criticism is that the mix is migrating from the highest-margin, longest-duration, most-defensible line (group LTD) into shorter-tail, lower-margin, more commoditized lines (STD, PFML, dental), that the flagship has lost share to eight of ten competitors for five years, and that per-share growth is being manufactured by the buyback rather than earned by the business: FY2025 aggregate adjusted operating income fell 11.5% while adjusted EPS fell only 3.7%, and Q1-26’s headline “+9.7% per-share growth” sits on +1.1% of actual dollars. Growth that exists only after dividing by a shrinking denominator is not growth.

6. Financial Quality

6.1 Benefit ratios — the core operating KPIs

Group disability, international, supplemental and dental are all deteriorating. Group life is the sole improver and management has disowned the Q1-26 print. Colonial is stable.

6.2 The GAAP-versus-adjusted wedge — this section is the report’s centerpiece

Management’s after-tax adjusted operating income: $917.9M (FY21) / $1,294.2M (FY22) / $1,513.6M (FY23) / $1,588.2M (FY24) / $1,406.0M (FY25); adjusted EPS $4.48 / $6.40 / $7.66 / $8.44 / $8.13.

The FY2025 reconciliation, verified to foot (after-tax, $M):

Are the adjustments legitimate? Item by item:

• Legitimate: net investment gains/losses (mark-to-market noise); the $82.0M pension annuity settlement (genuine, discrete de-risking); the FY2024 legal settlement.
• Soft: the $15.8M charitable contribution — discretionary spending, immaterial.
• Not legitimate: amortization of the cost of reinsurance. This is the contractual, predictable, escalating price of a deliberate strategy — $44.1M → $41.4M → $116.7M pre-tax, or $0.53 per diluted share and rising. Excluding it removes the price tag while keeping the benefit.
• Not legitimate over a cycle: reserve assumption updates. They appear in five of five years and they do not average to zero: pre-tax −$177.2M (FY23) / +$357.4M (FY24) / −$478.5M (FY25) = −$99.4M per year, or −$78.2M after tax. Two of three recent reviews were charges. A cost that recurs every year is an operating expense.
• Least legitimate: the Q1-2026 redefinition. Management now excludes the entire Closed Block segment from adjusted operating income, and simultaneously stopped adjusting for cost-of-reinsurance amortization “because the majority of these items are included in Closed Block results.” Prior periods were restated. The effect: Q1-2026 GAAP net income of $232.0M ($1.41) becomes adjusted $352.5M ($2.14) by excluding a $(116.5)M after-tax Closed Block loss. The reported “+9.7% per-share growth” is measured on a base that now omits the loss-making segment. This change was made in the exact quarter the Closed Block swung from +$8.0M to −$145.3M, and while it still carries $16.7bn of LTC reserves and consumes real statutory capital.

A second, deeper pattern the exclusions conceal. The consolidated assumption-update line looks near-neutral across five years (+$180M pre-tax cumulative), but it hides two persistent and opposite biases:

Unum US released reserves in all five years (+$231.7M FY21, +$170.8M FY22); Colonial in all five. LTC charged in three of five. Five years of over-reserving in the good businesses have been funding five years of under-reserving in the run-off block — and management excludes both legs, so neither is visible in the headline. An analyst put this directly to management on the Q3-25 call: “this is the fifth straight year that you’ve had a positive assumption review in that business.”

The cleanest comparable — GAAP EPS with assumption updates stripped from both sides: FY23 $7.20 / FY24 $7.96 / FY25 $6.45.

Defensible core earning power: $1,406.0M less cost-of-reinsurance amortization ($92.2M) less the normalized assumption drag ($78.2M) ≈ $1,236M ≈ $7.15/share; also removing non-contemporaneous reinsurance ≈ $1,212M ≈ $7.01/share. Management’s $8.13 overstates sustainable earning power by roughly 12–14%.

6.3 Book value quality — the AOCI mirage

69% of the $2,384M increase in GAAP equity from FY22 to FY25 is AOCI reversal, not earnings. GAAP BVPS rose 51.9%; ex-AOCI BVPS rose 26.6%. $1,795.5M of negative AOCI remains — $11.17 per share, 16.5% of GAAP book. If rates fall further, GAAP book will rise roughly 16% with zero economic improvement. Critically: ex-AOCI equity FELL 4.1% in FY2025 while GAAP equity rose 1.4% — they moved in opposite directions, and ex-AOCI BVPS rose only because buybacks cut the share count 7.2%.

ROE computed from the filings:

FY2024’s flattering 17.3% GAAP ROE is an AOCI-suppressed denominator plus a +$282.6M assumption benefit — not a real 17% business.

6.4 The LTC block, in detail

• Gross PV of future benefits $24,420.6M; PV of net premiums $7,732.7M; net LFPB $16,687.9M; less reinsurance recoverable $3,369.4M → $13,322.6M retained after Fortitude I.
• Undiscounted future benefits $59,239.1M against undiscounted future gross premiums $12,601.0M — a 4.7:1 ratio. Accretion rate 5.6%; current discount rate 5.5%; duration 14.7 years; persistency 95.8%.
• ~835,000 insureds (752k group / 83k individual) as of 9/30/2025; average issue date 2004; ~98% indemnity-based; average attained age 59 (active life) / 83 (disabled life).
• The Q3-2025 charge, itemized (management’s own decomposition): incidence assumption raised +$300M; disabled-claim mortality raised −$200M; full removal of the morbidity and mortality improvement assumptions +$850M; discontinuation of new-employee enrollment on group cases +$200M; expanded premium-rate-increase program −$525M. Net ~$643.1M. Only ~$100M net is observed experience. The $850M was explicitly attributed to modeling uncertainty: “the trend hasn’t really fully reemerged… it becomes very tough to model it.”
• The analyst pushback deserves recording. Tom Gallagher (Evercore) put the circularity to management directly: “was there anything in the experience… that would warrant these long-term assumption changes… Or was this more due to future uncertainty and prudence? …if I’m the regulator seeing this request, is this viewed as — if a management team is just becoming more prudent, are they still likely to approve it?” CFO Zabel conceded “there’s a lot of actuarial judgment around this,” and confirmed on the record that the −$525M rate-increase credit flows from the discretionary assumption changes, not from experience. In other words: a discretionary reserve increase generated a discretionary reserve offset, and the regulator has to agree.
• The net premium ratio is the single most important number: 93.5% (2023) → 94.6% (2024) → 97.5% (2025) → 97.6% (Q1-26). Not capped in aggregate — but the 10-K explicitly cites “the impact of capped cohorts” driving higher benefit expense, and Zabel conceded on the Q3-25 call that “the more capped cohorts you get, probably the more volatility you’re going to see.” Unum does not disclose what share of cohorts is already capped. That is the most important undisclosed number in the filing. Capped cohorts cost −$436.7M pre-tax in FY2025, and capped cohorts have now also appeared in Colonial Life cancer/critical-illness (−$11.5M).
• FY2025 experience was adverse even excluding assumptions: a −$119.6M variance on the net-premium leg.
• Management’s own sensitivities (PV gross LFPB, net of reinsurance, unfavorable/favorable): lapse + mortality ±7% → $430M/$400M; claim incidence ±3% → $370M/$370M; claim resolutions ±2% → $290M/$280M; future unapproved rate increases ±10% → $140M/$140M, implying a ~$1.4bn embedded assumed-rate-increase asset if roughly linear (INTERPRETATION). Only ~15% of the rate-increase program embedded in the 2025 best estimate has been achieved.
• Fairwind captive “protections” fell $2.6bn → $2.0bn in 2025, and management confirmed the entire decline was best-estimate reserve margin — excess capital untouched.

6.5 Investment portfolio — clean, and not where the risk is

Total investments $43,520.8M. Fixed maturities $33,056.6M (75.9%), mortgage loans $2,109.5M, policy loans $3,668.1M, other long-term $1,670.4M, short-term $3,016.2M.

• Below-investment-grade is just 2.8% of the portfolio (down from 3.1%); the credit-loss allowance on a $34.7bn bond book is $5.9M. The only material credit events in three years were two communications issuers in 2025 (~$13.4M and ~$14.0M gross).
• Office CRE is a non-issue: office is 14.9% of a $2,109.5M mortgage book = 0.7% of total investments; 80.6% at LTV ≤65%; 94% internally rated BBB+ or better.
• Net unrealized loss $1,690.6M (improved from $2,640.0M) — rate-driven, not credit.
• Alternatives: private-equity partnerships $1,456.3M with $756.7M unfunded, marked at NAV on a one-quarter lag; NII contribution $78.1M / $103.1M / $91.2M — the “miscellaneous investment income” swing factor, and now moved below the adjusted-operating line for 2026.
• ALM: portfolio duration is not disclosed — but $2,603.0M of the $3,818.2M derivative notional (68%) hedges LTC reinvestment risk, implying management itself sees the block as under-matched at the long end.

6.6 Capital, leverage, and the cash-flow constraint

Debt $3,767.6M; debt/capital 25.3% GAAP, 22.6% ex-AOCI, 25.4% including $643.8M of FHLB advances that management’s ratio excludes. Interest coverage fell to 5.5x GAAP from 12.2x (9.5x on adjusted). No maturity before 2028; the ladder runs to 2058. The November 2025 refinancing stepped the coupon up 137bp (3.875% → 5.250%). Ratings: senior bbb+/BBB/Baa2/BBB; financial strength A/A/A2/A; all stable, no changes despite the charge. RBC 440%.

The statutory deterioration is the under-reported story. Traditional US life subsidiary net income fell $1,322.4M → $634.0M (−52%); combined statutory surplus fell $139.1M to $3,770.6M; the Fairwind captive’s net income collapsed $662.2M → $52.5M. Fairwind’s $2,003.3M surplus includes a ~$185M Vermont permitted-practice benefit that does not exist under NAIC statutory accounting (down from $334M in 2024).

For an insurer, “free cash flow” is subsidiary dividends to the holding company, not GAAP operating cash flow — and GAAP OCF of $687.7M in FY2025 is unusable anyway, distorted by Fortitude Re moving $935.5M of cash out. Unum Group received $1,547.1M in subsidiary dividends in 2025, but $787.9M of that was extraordinary, including $630.0M from First Unum approved only after First Unum ceded 100% of its LTC book to Provident.

The constraint that matters: 2026 ordinary dividend capacity is $631.4M (down 54% from ~$1,383M), plus ~£125M — roughly $800M — against $1,316.5M returned in 2025 and a similar amount guided for 2026. The ~$500M annual gap is bridgeable for two to three years from $2,344.1M of holdco liquidity and further extraordinary approvals. But it means the buyback is partly funded from a stock of capital, not from flow.

6.7 Normalized earnings power

Assumptions: group disability to ~63%; group life to ~70% (versus 67.5% FY25 and 61.8% Q1-26); Colonial to ~50%; Closed Block to −$75M (range −$250M to +$50M — the widest band, worth ±$0.90 of EPS); private-equity NAV at its three-year $91M average; no assumption charge or benefit; 164M shares.

Verdict: normalized operating ROE ≈ 9.1% on ~$12.9bn of ex-AOCI equity — approximately the cost of equity for a BBB-rated life insurer carrying a $16.7bn LTC tail. The asset side is clean; essentially every risk in this company is on the liability side. Economics do not improve with scale here, because the binding constraint is not expense efficiency — Unum’s expense ratio has been flat at 22.5%/22.5%/22.6% for three years — but the run-off block’s consumption of capital.

7. Capital Allocation

7.1 The buyback — an excellent record whose conditions have expired

60.3M shares — 19% of the company — retired at a blended $54.66 against today’s $89.05: roughly $2.07bn of value created. Nothing was bought in the 2020 panic. On the Marathon test this is close to a model program.

But 72.6% of all dollars were deployed in 2024–Q1’26 at an average $67.94, versus $34.09 for 2019–23 — and the P/adjusted-book paid has climbed monotonically from 0.45x to 0.96x. The excellence of this record was a function of the price at which capital was deployed, and that price has normalized. Historic buyback skill is not a permanent attribute; it came from buying a de-rated block, and the de-rating is gone.

Is management buying at the high? Not yet, and the framing must be precise. Q1-26’s $74.52 average reconciles to the tape — UNM only broke $89–93 after 2026-02-27 — and management characterized Q1 as an explicit pull-forward, not an increase (“on track to repurchase $1 billion of stock this year”). That leaves ~$601M across Q2–Q4 at roughly $200M a quarter, at ~1.13x adjusted book and the 98.4th percentile of the stock’s own ten-year P/B and P/E range. The Q2 10-Q’s monthly repurchase table (~July 29) is the single artifact that settles whether this program has just made its first expensive purchase.

The “debt-funded buyback” hypothesis is wrong. Net cash from debt was +$21.0M (2025) and +$41.6M (2024) — pure refinancing. The rise in long-term debt to $3,767.6M is largely non-cash accretion on the $400M P-Caps 2041 notes recognized at a $273.5M fair value in October 2024. The real funding stress is the Section 6.6 dividend-capacity gap, not leverage.

Asked directly why he was buying at $89, CEO Richard McKenney called it “attractive prices” while offering no multiple, no book reference, and no intrinsic-value frame — and when pressed, gave a capital-availability rationale (“we sit in excess capital”), not a price one. That reasoning justifies buying at any level.

7.2 The LTC risk-transfer decisions

Fortitude I (announced 2025-02-27, closed July 2025, effective 2025-01-01): ~21% of Closed Block LTC future policy benefits and ~15% of Unum US individual disability FPB. Consideration: $953.5M cash + $3,230.1M of securities + $47.1M accrued income = $4,230.7M against $3,620.5M of ceded GAAP reserves — a ~17% premium. Ceding commission $442.3M. Cost of reinsurance $848.2M (LTC), partly offset by a $145.9M deferred gain on the IDI leg after a $100.3M DAC write-off; $46.8M pre-tax net investment loss. Collateralized trust established for Unum’s benefit; Fortitude retroceded 100% of biometric risk to an unnamed highly-rated global reinsurer. Net capital benefit: ~$100M (−$200M on LTC, +$300M on IDI). Recurring drag now $116.7M pre-tax = $0.53 per diluted share.

Fortitude II (announced 2026-07-02, 8-K filed 2026-07-06, effective 2026-04-01, closing during 2026): ~50,000 individual LTC policies, average attained age 76, $3.8bn of statutory reserves / ~$4.5bn best-estimate, with ~$5.7bn of assets transferred — a ~27% premium over best estimate and ~50% over statutory. Equals 26% of total LTC statutory reserves and 52% of individual LTC reserves. Fortitude again retrocedes biometric risk; Unum retains administration and provides a $125M (NPV-capped) experience-volatility cover to the retrocessionaire for a $5M fee, settled every five years and trust-secured. Funded from Fairwind excess capital, holdco liquidity, and financing against future tax benefits. Post-close targets: holdco liquidity $1.5–2.0bn, leverage ~25%, RBC 400–425%.

Cumulatively: >$7bn of LTC reserves ceded, ~40% of the statutory block, in eighteen months. That is real execution and the market was right to reward it.

Three things cut against it. First, the price did not improve on the second bite — ~17% then ~27% over carrying reserves — despite eighteen additional months of rate-increase approvals. Unum is a price-taker in a market roughly one-to-two balance sheets deep; the Manulife/Global Atlantic deal required the identical workaround, and S&P called the structure “still a rarity.” Second, the GAAP loss on Fortitude II is undisclosed and is the largest known unquantified hit to 2026 earnings — the ~$1.9bn gap between assets transferred and statutory reserves implies a charge plausibly at or above the $848.2M booked on the smaller first deal (INTERPRETATION, labeled as estimate; expect quantification in the Q3-2026 10-Q). Third, and decisively for the thesis: both deals were individual LTC. The ~$11.0bn residual is ~70% group LTC — the category with worse rate-increase economics (requires plan-sponsor cooperation, higher shock lapse and benefit-reduction elections), no observed buyer in any completed industry transaction, and the block Unum is currently flagging as adverse (Q1-26 Closed Block loss “driven primarily by group policy terminations and claim incidence in the long-term care product line”).

7.3 M&A — the strongest item in the case

Essentially none, for a decade. One $145.4M outflow in FY2018 (Unum Poland; $12.6M of premium). Starmount (2016) immaterial. No acquisition line FY2019–FY2025. All debt movement was refinancing. In an industry with at best narrow barriers, where peers have been buying scale aggressively (Meiji Yasuda/StanCorp three times in four years; Sun Life’s $2.6bn DentaQuest purchase, now impaired; Voya’s Benefitfocus, described by Bloomberg as “ill-fated”), not empire-building is the correct Greenwald answer and management has held that line. Credit where due.

7.4 Compensation and incentive alignment — the first-order governance finding

From the DEF 14A filed 2026-04-09.

Annual incentive metrics: After-Tax Adjusted Operating EPS 50%, core Premium Income 20%, Sales 10%, Customer Experience 10%, Adjusted Other Operating Expense Ratio 10%.

The standing exclusions, quoted verbatim from the proxy, include: “The impact of any unplanned acquisitions, divestitures or block reinsurance transactions”; “Unplanned adjustments to the Closed Block”; “Unplanned reserve assumption updates”; and “…or stock repurchase or issuance.” The proxy confirms which were applied for 2025 — including “The effect of reserve assumption updates.”

The wedge, plainly: GAAP diluted EPS fell 54.9% to $4.27. The metric management is paid on was $8.13 — 90% higher. Payout was 76.4% of target.

The long-term plan carries the same scrub: “Adjusted Book Value is total stockholders’ equity adjusted to exclude AOCI; and was further adjusted to exclude… Net realized investment loss and other impacts from reinsurance transactions; Reserve assumption updates; Cost related to… stock repurchases…” The 2023 cycle paid out at 172.6% (adjusted-book-value ratio 129.0%, business results 143.8%, relative TSR at the 100th percentile), earning the CEO $7,767,000 in cash — over a 2023–25 window in which the LTC assumption line ran a cumulative −$837.1M.

FY2025 NEO awards (all at 76.4% of target): McKenney $2,154,995; Zabel $890,905; Pyne $620,750; Anderson $307,216; Iglesias $476,912. CEO SCT total $19,417,476; pay ratio 1-to-222. For 2026 the CEO’s annual-incentive target rose from 250% to 260% of salary and LTI from $11.3M to $11.651M.

The counterweights are real and should not be suppressed: the 76.4% is a genuine haircut from 100%; the 172.6% was market-validated by a 100th-percentile relative TSR against a named peer group; excluding buyback effects from an EPS metric is shareholder-friendly, not the reverse; and 2026 adds PSUs with new targets. Governance hygiene is mostly good: McKenney holds 67.0x salary ($76.0M) against a 6x requirement; single share class; no pledging; anti-hedging; two clawback policies (Rule 10D-1 plus a broader misconduct policy requiring no restatement). Say-on-pay support was 96.1%, up from 92% — support rose in the year earnings halved.

The residual objection survives all of that: there is no GAAP, statutory, ROE, or reserve-adequacy measure anywhere in the incentive plan. Relative TSR is a ±20% modifier on a scrubbed base, not a gate. Management can lose $1.0bn of GAAP net income and $837M cumulatively on LTC assumptions and still collect 172.6% of a long-term award. If the metric excludes the reserve charges, the metric creates an incentive to under-reserve and to classify strengthening as non-operating. That is a structural defect in plan design, not an accusation about any individual.

Aggregate insider ownership is under 1% (19 persons, 1,608,925 shares; the CEO 0.52%).

7.5 Management and the transition question

There is no CEO or CFO transition. A systematic scan of all 52 mirrored 8-Ks found zero Item 5.02 filings after 2024-01-04. Richard McKenney remains President & CEO (re-elected 2026-05-21); Steven Zabel remains CFO. The 2026 Form 3s and 8-Ks map to: Steve Jones becoming President of Colonial Life effective 2026-06-08 (Tim Arnold retiring after 41 years); Kristi Matus elected a director at the May 2026 annual meeting; Andrew Walker promoted to EVP Chief Customer Operations Officer (Item 8.01, a promotion, not a 5.02 departure). The May 2026 Form 4 batch is the routine annual director equity grant, all code A.

McKenney’s full tenure (CEO since 2015-05-21): TSR +261.4% / +12.20% annualized, versus the S&P 500’s +319.6% / +13.72% — underperforming the index by 1.51 points annually, while beating a peer group of nine (3rd, median +10.14%) that contains Lincoln at +0.72%. The path is the story: +76.6% to January 2018, then −81.1% to March 2020 (the LTC charge plus COVID — corroborated by FactorsToday’s −0.8107 lifetime maximum drawdown, the worst in the stock’s forty-year history and entirely inside this tenure), then +979.8%. Ex-AOCI BVPS compounded 9.1% over the tenure.

7.6 Insider transactions

Parsed from 112 raw ownership XML filings (avoiding the rendered-HTML parsing trap), 219 itemized lines: 102 code F, 73 code A, 34 code S, 7 code G.

Zero code-P open-market purchases in thirty months. Fifteen sales across 2025–26 totaling $9,791,103, of which fourteen of fifteen were discretionary (only one was 10b5-1-planned). The largest: McKenney sold 50,000 shares in three tranches on a single day, 2025-03-03, for $4,099,453, discretionary. CFO Zabel sold 7,500 shares at $76.38 on 2025-11-19, three weeks after the Q3 charge print.

But the “selling into the all-time high” premise does not hold and should not be asserted. Every sale executed between $72.47 and $82.75 — all below today’s $89.05, the highest at $82.75. Insiders sold on the way down and through the trough, and have been absent from the rally. The sharper and fairer indictment is the other side: at $72–73 in February–March 2026, with the charge fully public and the stock ~20% below spot, not one insider bought. At $9.8M against a $14.3bn market cap the scale is routine diversification; the signal is composition, not magnitude, and it should not be over-weighted.

7.7 Verdict

Qualified yes on execution; no on incentive design; and the marginal dollar is the weakest it has been. The affirmative case is strong: 19% of the company retired at 0.45–0.96x book, nothing bought in the panic, no empire-building for a decade, debt refinanced rather than levered, ~40% of LTC statutory reserves transferred, and a genuinely excellent, well-timed exit from medical stop-loss immediately before that market’s blowup.

Against it: the LTC exits are expensive and did not get cheaper, with $0.53/share of permanent amortization drag and reinsurance-recoverable concentration in essentially two counterparties; the return program has outrun its cash generation ($1,311.4M returned on ~$555.4M of FCF, bridged by extraordinary dividends, with RBC being deliberately spent down and no committed timeline); and the incentive architecture does not measure what has determined shareholder outcomes. The forward judgment matters more than the record: this program’s excellence came from buying a de-rated block, and the de-rating is gone.

8. Changes and Headwinds — Last Two Years

2024-07 — Medical stop-loss sold to Amynta Group (closed Q3-2024; terms undisclosed). Unum stopped actively marketing stop-loss in Q3-2024. Immediately preceded the industry’s loss-ratio blowup to a record ~91%. Strengthens the thesis on management quality; the run-off has depressed reported group disability premium for seven quarters.

2024-11 — FY2024 peak. GAAP EPS $9.46, ROE 14.6% — both materially flattered by a +$357.4M pre-tax assumption benefit that reversed the following year. Neutral-to-negative: the peak was not what it appeared.

2025-02-27 / 2025-07 — Fortitude I ($3.4bn statutory LTC + ~$120m IDI premium ceded; $848.2M cost of reinsurance). Strengthens: real de-risking. Weakens: expensive, and creates a permanent $0.53/share drag.

2025-07-30 — Q2-25 miss, −12.2%, the worst session of the five years (adjusted EPS $2.07 vs $2.23 consensus).

2025-11-03 — Q3-25: the $478.5M pre-tax / $377.8M after-tax reserve charge, of which Closed Block LTC was −$643.1M against +$162M of core releases. GAAP net income $39.7M vs $645.7M. The stock rose 7.2%. Weakens materially: third LTC charge in four reviews; NPR to 97.5%; and the discretionary $850M improvement-assumption removal, offset by a −$525M discretionary rate-increase credit, invited direct analyst challenge on circularity.

2025-11-14 — $300M of 5.250% senior notes due 2035 issued (refinancing; coupon stepped up 137bp). AM Best affirmed “bbb+”/stable — no adverse ratings action anywhere despite the charge.

2025-12-04 — new $1.0bn buyback authorization effective 2026-01-01, the first without an expiry.

2026-02-01 — Group LTC closed to new enrollments on existing cases. Management: “eliminated tail-risk.” Q1-26 saw 7% of cases terminate, ~30,000 net lives, with a statutory reserve release of “less than $100 million.” Strengthens structurally; costs premium and produced the group-termination losses now hitting the Closed Block.

2026-02-05 — FY2026 guidance and the metric redefinition. Closed Block (and the volatile alternatives portfolio) moved below the adjusted-operating line; FY2025 base restated $8.13 → $7.93; guidance $8.60–$8.90. FY2025 statutory earnings came in at $1.1bn against a $1.3–1.6bn guide (a 15–30% miss), and 2026 resets to $1.2–1.4bn. Weakens: the base was lowered in the same breath as the growth rate was declared.

2026-04-28 — Q1-2026. Net income $232.0M; Closed Block −$145.3M adjusted operating loss (from +$8.0M); NPR 97.6%; group disability BR to 63.7%; group life BR 61.8% (“an anomaly”); supplemental & voluntary AOI −17.4%; International AOI −25% versus its own outlook. Weakens: the headline beat is carried by mortality luck in one line while three others deteriorate.

2026-05-21 — Dividend raised ~10% to $0.505/quarter ($2.02 annualized).

2026-07-02 / 07-06 — Fortitude II ($3.8bn statutory / ~$4.5bn best-estimate; ~$5.7bn of assets). The stock fell 3.1%. Strengthens: takes cumulative LTC statutory de-risking to ~40%. Weakens: ~27% premium, a large undisclosed GAAP loss to come, and it exhausts the individual block — the ~$11bn residual is ~70% group LTC with no observed buyer.

Verdict: On balance these developments weaken the thesis relative to the price. The genuine strengthening — stop-loss exit, two LTC cessions, group LTC closed to new entrants — is real, is done, and is in the stock. The weakening — a third LTC charge, an NPR at 97.6%, benefit ratios past cycle peak across four lines and two geographies, a 54% cut to dividend capacity, a 15–30% statutory earnings miss, and a headline metric redefined to exclude the problem — is ongoing and is not in the stock.

9. Risk Analysis

Catastrophic-loss risk: low in the conventional sense — Unum’s asset side is clean, it has no maturity before 2028, RBC is 440%, and all four rating agencies are stable. The tail risk is not solvency; it is a multi-year sequence of reserve strengthenings on a block running to peak claims around 2041, which is precisely what produced the −81.1% drawdown of 2018–2020 and the comparable episodes of 2002–03 and 2008–09. That has happened twice before in this company’s history on this same liability.

10. Valuation Discussion

No price target and no recommendation appear in this section. What follows is embedded-expectations analysis and scenario work.

10.1 Where the stock sits

At $89.05: 1.31x GAAP book ($67.76 BVPS) but 1.13x ex-AOCI book ($78.93); 19.3x charge-distorted TTM GAAP EPS; 10.9x management’s FY2025 adjusted $8.13; 12.4x the normalized ~$7.20; 2.3% dividend yield. On the stock’s own ~10-year history: P/E 99.3rd percentile, P/B 99.3rd, P/S 96.8th, composite 98.4th.

The P/E percentile should be discounted — TTM EPS of $4.60 is charge-distorted, so 19.3x is not the right lens. P/B is the clean signal, and it is unambiguous. Unum’s own post-LDTI ex-AOCI series: 0.67x (FY22) → 0.67x (FY23) → 0.97x (FY24) → 0.99x (FY25) → 1.13x now. The discount closed during FY2024. Everything since has been a premium to book.

10.2 Sum-of-the-parts — what is the market paying for the Closed Block?

Built on an equity basis (allocated equity sums to total adjusted equity, and Corporate’s after-tax loss already carries the interest on the $3,767.6M of debt, so adding back debt would double-count). Normalized core after-tax AOI $1,384.1M; Corporate normalized −$146.2M capitalized at 10x = −$1,462M. The identity reduces to implied Closed Block value = $15,776.8M − core value, against a $14,314.8M market cap.

The SOTP has exactly one degree of freedom — core multiple and Closed Block value are perfectly substitutable — so what it establishes is a joint constraint, not a point estimate: at $89.05 you cannot simultaneously believe the core deserves 13x or more AND that the Closed Block is worth anything positive. Pick one.

Bracketing what the block should be worth: status quo (0.8% ROE against a 9.5% cost of equity) ≈ +$0.4bn; a clean full exit at Fortitude-II economics (~$2.4–3.5bn of cost against $4.83bn of capital released) ≈ +$1.3bn to +$2.4bn; adverse (NPR caps, sensitivities stacking to ~$1.1bn, plus impairment of half the ~$1.4bn rate-increase asset) ≈ −$1.0bn to −$2.5bn. The market’s implied ~$0 to +$552M sits between status-quo and mildly adverse — conservative, not euphoric, and notably not pricing a group-LTC exit at all.

External validation of the sign: Genworth’s $3,869M market capitalization is ~27% below the $5,286M market value of its 81.6% Enact stake alone — the market marks a large LTC run-off block at roughly −$1.4bn. Genworth has no core engine, so this is not a fair value for Unum’s block; read it only as evidence that positive marks on LTC run-off are not what the market does.

Caveat that must survive into any use of this analysis: the core’s 20.5% ROE rests on a management-constructed “allocated equity” denominator. The earnings ($1,384M) are robust; the ROE and the P/B-of-allocated-equity are soft. Lead with the P/E lens.

10.3 Embedded expectations

Cost of equity, derived rather than asserted. CAPM on beta 0.592 gives 7.3% — reject as a point estimate, because beta measures covariance, not a fat-tailed twenty-year liability, and roughly 70% of UNM’s return variance is idiosyncratic. Bond-yield-plus-premium (Baa2/BBB, cost of debt ~5.7%, plus 350–450bp) gives 9.2–10.2%. Market-revealed: a 12-year mean ROE of 9.33% against a stock that traded 0.41–0.78x book in every year from 2018 to 2023. Adopt 9.5% central, 9.0–10.0% band — noting that beta argues for 7.3% and the −81.1% lifetime drawdown argues for 11%+.

Solving the justified-multiple identity P/B = (ROE − g)/(COE − g) at the traded 1.128x ex-AOCI:

The cleanest statement in this article: against a ~$12.9bn ex-AOCI equity base, a normalized 9.14% ROE produces $1,179M, or $7.20 EPS; the market-implied 10.33% produces $1,332M, or $8.13 EPS. The market is underwriting precisely management’s as-reported FY2025 adjusted operating EPS — the number two independent workstreams conclude overstates sustainable earning power by 12–14%. The gap is 119bp of ROE, $153M after tax, $0.93 per share.

Decomposing the $10.12 per share ($1,627M) premium over ex-AOCI book:

• (a) LTC de-risking optionality: ~zero. The implied block value is approximately its status-quo value. The market is not paying for a successful exit.
• (b) Core re-rating: essentially all of it. At SOTP break-even the core carries the entire $15,777M, or 2.18x its allocated equity. This is the load-bearing assumption in the stock.
• © The buyback: contributes nothing to a justified premium. At ROE ≈ COE the justified P/B is nearly insensitive to growth (0.92x / 0.94x / 0.93x / 0.91x at g = 2/3/4/5%). A buyback cannot create value for a company earning its cost of capital — it can only transfer it, adversely so above intrinsic value. Management bought at $74.39 (FY25) and $74.52 (Q1-26); today is 19–20% higher. The funding is also fragile.
• (d) Factor and cohort beta: material. DividendYield (+0.42) and Value (+0.21) are Unum’s two largest style loadings and are the two hottest factors of the trailing year, both at roughly +1.7 standard deviations, while Financials is the most out-of-favor factor on the board (−11.7%, z −1.58).

The reconciliation, and the single question in the stock:

That $14.86 is the present value the market assigns to roughly half of the Closed Block’s $4.83bn of allocated capital being freed and redeployed at core returns. So the question is not “cheap or expensive.” It is whether that $4.83bn is surplus capital temporarily trapped — in which case $89.05 is fair to cheap — or permanently impaired capital consumed by a liability running to peak claims around 2041, in which case $89.05 is 13–20% too high.

10.4 Peer comparison

All prices at 2026-07-17 close; balance sheets at 2026-03-31. P/B recomputed from raw common equity ÷ actual shares outstanding — the aggregator’s own book-value-per-share and price-to-book fields were rejected because their share counts do not tie (UNM’s implied 136M against 160.7M actual).

* Distorted by denominator artifacts (VOYA’s tiny common equity, CNO’s LDTI transition, CRBG’s pre-IPO carve-out, EQH’s negative common equity). † IFRS-17 and CAD-reported; no ex-AOCI figure computable. MetLife’s ~$3.2bn of preferred is not separately broken out, making that row the least reliable.

Against its own factor cohort, UNM is the cheapest name on ex-AOCI book (1.13x < PRU 1.17 < MET 1.30 < CNO 1.35 < VOYA 1.38 < RGA 1.44 < EQH 2.52). A regression of ex-AOCI P/B on 5-year ROE across the eight usable names gives P/B = 0.334 + 0.0985 × ROE(%), fitting UNM at 1.23x on its normalized 9.1% ROE — so it screens roughly 8% below the line, not rich.

Three qualifications gut most of that comfort. First, UNM has the lowest normalized ROE in the group, so the cheapest multiple is partly earned. Second, the cohort itself is expensive against its own history — only LNC (55.1) and CRBG (66.7) sit below the 70th percentile of their own ranges. Third, and most important, the cohort carries live M&A premia UNM has not earned: VOYA is in play (activist TOMS Capital demanding a strategic review, Principal Financial having expressed interest, the stock at a multi-year high) and EQH/CRBG sit in an announced merger. UNM has no deal and no activist, yet sits at the 98.4th percentile. It may be marked to a transaction that has not been proposed.

And the regression cannot bear weight: LNC sits 0.9 turns below the line (0.59x on a 13.4% ROE) and AFL 0.9 turns above (2.26x on 9.7%) — both legacy-block life insurers. Dispersion at any given ROE is ±80% of the fitted value. What actually drives a life insurer’s P/B is confidence in the liability, which is precisely what is in dispute here. LNC is the empirical bear template; AFL is not a usable bull template — Aflac currently trades at close to the richest multiple in its modern history.

10.5 Scenarios

All anchored on ~$7.00–7.20 normalized EPS and $78.93 ex-AOCI BVPS — not $8.13, not the restated $7.93, not the $8.60–8.90 guide. Both lenses (multiple × normalized EPS, and P/B × book) are required to converge. These are conditional ranges, not price targets.

BEAR — implies roughly $50–65. The NPR reaches the 100% cap (97.6% now, drifting ~1.6pp/yr, ~18 months of headroom — this requires continuation, not acceleration); the capped block becomes one-directional; group LTC (~$7.7bn) proves untradeable; half the ~$1.4bn rate-increase asset is impaired; the Closed Block runs at −$250M/yr; core benefit ratios move past mid-cycle as the labor market softens; the buyback is cut to ~$475M to fit dividend capacity. This implies $1.0–1.5bn of pre-tax strengthening, taking ex-AOCI book to ~$71–74 and EPS to $6.00–6.30 at a 0.70–0.90x multiple — where UNM itself traded at 0.67x as recently as FY2023, and where LNC trades today at 0.59x.

BASE — implies roughly $75–90. The NPR drifts to ~99% without capping; Fortitude II closes absorbing a ~$1bn GAAP charge (estimate); the Closed Block runs at −$75M; the core operates at mid-cycle benefit ratios; the buyback runs $700–800M funded partly from holdco liquidity; EPS is $7.00–7.20 with aggregate earnings flat and all per-share growth coming from the share count. ROE of 9.1% against a 9.5% COE — and the justified multiple is almost insensitive to growth, which is the base case. $89.05 sits at the top of this band.

BULL — implies roughly $102–118. A group-LTC transfer is executed at a premium no worse than the ~25–27% of Fortitude II (industry deal sizes have scaled from $1bn to $4.4bn+, buyers have moved from average-attained-age-83 blocks to age-75 blocks, and the bifurcated structure is now standardized); this releases $4.0–4.8bn against $2.5–3.0bn of cost for a net $1.3–1.8bn redeployed; EPS moves to ~$7.55 then ~$8.50; the core re-rates to 12–13.5x.

Probability-weighted, the skew at $89.05 is unfavorable — because of an asymmetry in what each case requires. The bear requires only the continuation of two already-observed trends: an NPR that has moved 93.5%→97.6% with 2.4 points left, and a group block that has already printed −$145.3M. The bull requires an unprecedented event — a group-LTC bid, with zero instances across seven completed industry transactions, at a price that worsened between Unum’s own two deals (17% → 27%) despite eighteen months of additional rate approvals. The base case is modal and brackets the current price at its ceiling.

Sensitivity — three variables move the answer. The cost of equity is the most fragile input: at 8.5% the gap closes entirely and the stock is fair. The Closed Block normalization (−$250M to +$50M ≈ ±$0.90 of EPS ≈ ±$10 of value) is a wider band than the distance to either tail. And each turn of the core multiple is worth ~$1.38bn, or ~$8.60 per share, of implied residual.

10.6 What the market is underwriting correctly, and incorrectly

Correctly: the core is a good business and is not overpaid for (~11.4x at SOTP break-even is barely above Hartford and far below Aflac); management is genuinely executing the de-risking, and the skeptics who said the block market was closed were wrong; solvency and financing risk are correctly priced low (nothing matures before 2028, RBC 440%, $2.34bn of holdco liquidity, all four agencies stable); the market correctly trades on ex-AOCI book and ignores the AOCI mirage; and it correctly prices essentially no M&A risk.

Incorrectly, most important first:

1. It capitalizes $8.13 when the evidence supports ~$7.20 — 119bp of ROE, $0.93 per share. Reserve assumption updates have been recurring and net negative, and five straight years of core reserve releases are being treated as margin rather than as conservatism unwinding.
2. It capitalizes a buyback that is not funded from flow — roughly $500M a year ahead of ordinary dividend capacity, bridged from a stock of capital — and, at ROE ≈ COE, a buyback cannot create value at all.
3. It treats “two deals done” as evidence the third will happen, when the residual is categorically different: 70% group LTC, no observed buyer, and a demonstrated price-taker.
4. It does not price the convexity at a 97.6% NPR — which is a change in the shape of the payoff, not its level. Q1-26’s −$145.3M was the first observation of the new regime, and the stock made an all-time high ten weeks later.
5. It mistakes factor and cohort beta for company improvement — stretched Value/DividendYield style beta against an out-of-favor sector, in a cohort carrying transaction premia UNM has not earned.

The honest counterweight, stated plainly: UNM screens ~8% below its cohort’s regression line, is the cheapest name in that cohort on ex-AOCI book, and the SOTP shows the market marking the Closed Block at roughly zero — conservative, and corroborated by Genworth. The reconciliation with the 98.4th-percentile reading is the intellectual core of this article: Unum spent a decade at 0.4–0.9x book because the market treated the LTC block as unquantifiable. It now treats it as quantified and partially transferred. That re-rating was legitimate and is essentially complete. The remaining upside requires the group block to trade.

11. Variant Perception

11.1 Consensus

Sell-side opinion is genuinely split, not skewed — consistent with the observation that the multiple re-rated while fundamentals did not. The prevailing constructive view runs: a 20%+ ROE core franchise; ~40% of the LTC block de-risked in eighteen months with more to come; management guiding 8–12% EPS growth and a 15–17% adjusted ROE; ~100% of free cash flow returned; and a stock still under 11x adjusted earnings. The bear view is thinner in the market than it should be, and mostly reduces to “LTC is unknowable.”

11.2 The strongest bull case

Unum has done what the market said it could not: found a repeat counterparty, moved ~40% of the LTC statutory block in eighteen months, closed group LTC to new enrollments, and exited medical stop-loss immediately before that market’s worst year on record. The industry block-transfer market is deepening, not closing — deal sizes have scaled from ~$1bn to $4.4bn+, buyers have moved from mature blocks (average age 83) to younger ones (age 75), and the bifurcated asset-manager-plus-global-reinsurer structure is now standardized. If group LTC trades, ~$4.8bn of capital earning 0.8% is released into a franchise earning 20.5%, and the consolidated ROE roughly doubles. Meanwhile the core is #1 or #2 in its flagship line, Colonial Life out-margins Aflac US, the asset portfolio is genuinely clean (2.8% below-investment-grade, 0.7% office CRE), 19% of the company has been retired at an average 0.6x book, and the stock is still the cheapest name in its own factor cohort on ex-AOCI book, ~8% below the cohort regression line.

11.3 The strongest bear case

The re-rating is finished and was earned by de-risking the half of the block that had a bid. After Fortitude II, ~70% of the ~$11bn residual is group LTC, and there has never been a group-LTC transaction. The block’s net premium ratio is 97.6% with ~2.4 points to a cap that converts it into an instrument where only bad news gets recognized; capped cohorts already cost $436.7M pre-tax in FY2025 and have now appeared in Colonial Life as well. Only ~15% of the assumed rate-increase program — an implied ~$1.4bn embedded asset — has been achieved, and a large part of the FY2025 charge’s offset was a discretionary rate-increase credit that regulators must still grant, a circularity an Evercore analyst put to management directly. All three core tailwinds (life mortality, disability incidence, discount rates) peaked together in FY2024 and are reversing, with STD incidence — the leading indicator for LTD by construction — already rising. The flagship line has lost share to eight of ten competitors for five years. Statutory earnings missed guidance by 15–30% and 2026 ordinary dividend capacity fell 54% to $631M against ~$1.3bn of guided return. And management responded to all of this by redefining its headline metric to exclude the loss-making segment in the exact quarter that segment lost $145.3M, on a comp plan that already excludes reserve assumption updates, block reinsurance transactions, Closed Block adjustments, and buyback effects. At the 98.4th percentile of its own ten-year valuation range, none of that is in the price.

11.4 The 3–5 assumptions that matter most

1. Group LTC (~$7.7bn) can be risk-transferred at an acceptable price. No precedent exists. This is the hinge of the entire investment case.
2. The LTC net premium ratio stops short of the 100% cap. Requires the 2023–26 drift (~1.6pp/yr) to arrest within ~18 months.
3. Normalized earning power is ~$8.13, not ~$7.20. Requires reserve assumption updates to average to zero and cost-of-reinsurance amortization to be genuinely non-operating.
4. Group disability benefit ratios hold at 62–65% through a labor-market cycle. Never tested; the counter-cyclicality of disability claiming is well documented.
5. Capital return is sustainable at ~100% of FCF. Requires either continued extraordinary dividend approvals or a successful third transaction to restore capacity.

11.5 The factor-positioning read

The tape does not confirm a fundamental inflection. FactorsToday shows UNM carrying a zero Momentum loading — L1-zeroed out of both the base (12-factor) and full (44-factor) models — despite sitting 4% off an all-time high. Its actual exposures are DividendYield +0.42, Insurance industry +0.48, Financials sector +0.42, CreditRisk +0.37, Market +0.63, Value +0.21, Quality +0.13. Empirically this is a dividend-yield, credit-spread, low-volatility insurance carry name — not a quality compounder, and only marginally a value name now that the discount has closed.

The risk-adjusted record is a three-to-five year phenomenon only: 3-year Sharpe 0.99, 5-year 0.94, but 10-year 0.34 and lifetime 0.23, against an −81.1% lifetime maximum drawdown realized twice on legacy-block reserve failures. The 1-year Sharpe (0.43) has already halved versus the 3-year — return has decelerated, volatility has not. Idiosyncratic volatility is 20.7% against ~24.8% total, with R² of 0.46: roughly 70% of variance is stock-specific, and the two largest specific-return days in the window are both prints (−11.3% on 2025-07-30, +6.4% on 2025-11-04).

Two things follow (INTERPRETATION, regime-caveated). First, UNM’s two largest style loadings are precisely the two hottest factors of the past year, both at roughly +1.7 standard deviations — a material share of the re-rating is factor beta to a stretched yield/value regime rather than company-specific improvement. Second, the all-time high was achieved against its sector: Financials is the most out-of-favor factor on the board (z −1.58) and the Insurance industry factor is flat. This move is idiosyncratic, not a sector tide.

Framing verdict: a completed value re-rating now being held as an income/defensive carry trade — neither a momentum one-way street nor a falling knife. Mean-reversion risk therefore sits in two places: the regime (Value and DividendYield at ~+1.7z cannot compound indefinitely) and the tail (a −81% lifetime drawdown, twice realized on exactly this liability) — not in a crowded momentum unwind. The 3.1% decline on the Fortitude II announcement is the first datapoint suggesting the market may begin scrutinizing the price of de-risking rather than applauding the fact of it.

12. Fact vs. Interpretation

13. Open Questions

1. What share of LTC cohorts, by reserve, is already net-premium-ratio capped? Undisclosed, and decisive — it determines how much of the block is already in one-way loss-recognition. The single most important undisclosed number in the filing.
2. What GAAP loss will Fortitude II book at closing? Undisclosed; the ~$1.9bn asset-versus-statutory-reserve gap implies it is large. Expect quantification in the Q3-2026 10-Q.
3. Is there any market bid for group LTC at all? If not, ~$7.7bn is permanent rather than temporary, and the SOTP’s implied ~$0 mark on the block is generous rather than conservative.
4. Who is the unnamed retrocessionaire holding the biometric risk on both Fortitude transactions? Undisclosed on both deals; it is the true ultimate counterparty.
5. Will FY2025 comparatives be restated onto the Closed-Block-excluded basis in the FY2026 10-K, further flattering reported growth?
6. Did the buyback continue at $89? The Q2-26 10-Q’s monthly repurchase table (~2026-07-29) settles whether this program has made its first materially expensive purchase.
7. What is the portfolio’s duration and duration gap against a 14.7-year-duration LTC liability? Not disclosed; the fact that 68% of derivative notional hedges LTC reinvestment risk suggests management sees a gap.
8. How much of the FY2024 ~59% disability benefit ratio was discount-rate benefit versus true experience? Not disclosed — without it, the 62–64% guide cannot be independently verified.
9. Is falling persistency (LTD 93.3%→91.1%) deliberate pricing discipline or share loss? The industry-wide decline in new-sales premium-per-life argues for the latter.
10. Does the Fortitude II cession restore 2027 ordinary dividend capacity, or does it consume the Fairwind capital that was funding the gap?
11. Minor, unreconciled: the FY2025 10-K states Fortitude I cash transferred as both $953.5M and $935.5M in different sections.

14. What Must Be True

14.1 For the bull case

14.2 For the bear case

15. Source Appendix

See the accompanying source appendix for the full itemized list. Principal primary sources used throughout:

• Unum Group FY2025 Form 10-K (filed 2026-02-17) — https://www.sec.gov/Archives/edgar/data/5513/000000551326000008/unm-20251231.htm
• Unum Group Q1-2026 Form 10-Q (filed 2026-04-29) — https://www.sec.gov/Archives/edgar/data/5513/000000551326000044/unm-20260331.htm
• Unum Group DEF 14A (filed 2026-04-09) — https://www.sec.gov/Archives/edgar/data/5513/000000551326000026/unm-20260409.htm
• 8-K, Fortitude Re transaction II (filed 2026-07-06, event 2026-07-02) — https://www.sec.gov/Archives/edgar/data/5513/000000551326000069/unm-20260702.htm
• Q1-2026 statistical supplement (Ex-99.2, 2026-04-28) — https://www.sec.gov/Archives/edgar/data/5513/000000551326000041/unm03312026exhibit992.htm
• Q4-2025 statistical supplement and 2026 Outlook (Ex-99.2, Ex-99.3, 2026-02-05) — https://www.sec.gov/Archives/edgar/data/5513/000000551326000002/unm12312025exhibit992.htm
• Unum Group FY2025 Annual Report to Shareholders (2026-04-09) — https://www.sec.gov/Archives/edgar/data/5513/000000551326000030/unm-2025_ars.pdf
• Unum Group earnings-call transcripts — Q3-2025 (2025-11-04), Q4-2025 (2026-02-05), Q1-2026 (2026-04-29)
• Full 5-year SEC corpus (10-K ×5, 10-Q ×15, 8-K ×52, DEF 14A ×5, Form 3/4/5 ×112) mirrored locally
• Milliman, 2025 US Group Disability Market Survey (2025-11-24) and prior editions; LTC M&A survey (2026-02-03)
• Gen Re, 2025 US Group Term Life Market Survey (2026-05-28)
• LIMRA workplace benefits releases (2025–2026)
• NAIC 2024 Market Share Reports — Life & Fraternal; Accident & Health
• Oliver Wyman / Guy Carpenter, Stop Loss Market Update, Fall 2025
• Swiss Re Group Watch (UK group risk, 2025)
• FactorsToday factor model — stock loadings, leaderboard, related stocks, factor returns (2026-07-17)
• AZI price history and valuation-index percentiles (2026-07-17)
• ROIC.ai financial statements, ratios, valuation multiples, transcripts
• Peer primary filings: MET, HIG, LNC, PFG, AFL, PRU, VOYA, SLF, CNO, GNW, RGA, EQH, CRBG, and Japanese-parent disclosures for Tokio Marine (Delphi/Reliance) and Meiji Yasuda (StanCorp)

Sections 1–15 carry no investment recommendation and no price target. The Claude's Take block at the head of this article is a clearly-labeled exception representing the author’s own subjective view. This article is general information, not investment advice.

APPENDIX A — Standard Diligence Questionnaire

Unum Group (NYSE: UNM) · 2026-07-18

A standard diligence questionnaire, supplemental to the main analysis. Answers are labeled FACT / INTERPRETATION / ASSUMPTION where it matters, and apply the Competition Demystified and Capital Returns frameworks where they add insight. Where a question does not map to a group-benefits insurer’s business model, the correct sector analog is given.

General

What thoughtful questions have other investors asked about this company?

The best questions came from sell-side analysts on the Q3-2025 and Q1-2026 calls, and they are the right ones:

• Tom Gallagher (Evercore ISI), on circularity in the LTC charge: “was there anything in the experience… that would warrant these long-term assumption changes… Or was this more due to future uncertainty and prudence? …if I’m the regulator seeing this request, is this viewed as — if a management team is just becoming more prudent, are they still likely to approve it?” CFO Zabel conceded “there’s a lot of actuarial judgment around this,” and confirmed on the record that the −$525M rate-increase credit flows from the discretionary assumption changes rather than from observed experience. FACT. This is the sharpest question anyone has asked Unum: a discretionary reserve increase generated a discretionary reserve offset, and a regulator must ratify it.
• Michael Hurwitz (Dowling), on earnings quality: “this is the fifth straight year that you’ve had a positive assumption review in that business” — pointing at the Unum US releases that have been funding the LTC charges. FACT.
• Suneet Kamath / Joshua Shanker (BofA), on the group-LTC enrollment freeze: “So you’re foregoing positive cash flows and business because you simply want to make the book smaller?” CEO McKenney: “That is correct.” FACT.
• Wes Carmichael, on capped cohorts: drew from Zabel the concession that “the more capped cohorts you get, probably the more volatility you’re going to see.” FACT.
• Michael Burdis (Raymond James) offered the bull framing on the Q3-25 charge — no cash impact, more future cash, fewer future charges — which Zabel accepted verbatim. FACT. This exchange is the best available explanation for why a 94% GAAP earnings collapse produced a +7.2% stock move.

The question nobody has asked, and should: what share of LTC cohorts, by reserve, is already at the 100% net-premium-ratio cap? Management discloses the aggregate ratio (97.6%) and concedes capped cohorts exist and cost $436.7M pre-tax in FY2025, but never quantifies the proportion. It determines how much of the block is already in one-way loss recognition. This is the single most important undisclosed number in the filing.

The second question nobody has asked: what GAAP loss will Fortitude II book at closing? The ~$1.9bn gap between assets transferred (~$5.7bn) and statutory reserves ($3.8bn) implies a large charge. Not in the press release, not in the 8-K.

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low?

A cyclical high on the operating side, masked by a cyclical low in reported GAAP. This distinction is the crux and it is easy to get backwards. GAAP EPS of $4.27 in FY2025 looks like a trough — and on GAAP it is. But the underlying operating earnings sit near a peak, because three independent tailwinds crested together in FY2024 and are now reversing:

1. Group life mortality. Benefit ratio 78.7% (FY22) → 66.3% (FY24) → 61.8% (Q1-26), against management’s own 68–72% guide and a stated historical average “in that high 60% range.” The CFO called Q1-26 “really an anomaly.” FACT.
2. Disability incidence and recovery. Benefit ratio bottomed at 59.0% (FY24) and has risen to 62.4% → 63.7% (Q1-26). Management guides to 62–64% near-term and ~65% “over the next few cycles.” FACT. This one has already turned.
3. Discount rates. The disclosed “Effect of Change in Discount Rate Assumptions on the LFPB” fell from $1,185.4M ($6.64/share) at 12/31/24 to $929.9M ($5.61/share) at 12/31/25 — ~$255M of rate-driven reserve relief handed back in one year. FACT.

INTERPRETATION: normalized operating EPS is ~$7.00–7.20, below management’s reported $8.13, and materially below the $8.60–8.90 FY2026 guide.

Driven by the external environment or internal actions?

Both, in a specific ratio. The favorable 2021–24 margin was overwhelmingly external — post-COVID mortality normalization, a benign labor market suppressing disability incidence, and a rate cycle that lowered new-claim reserves. Management’s genuine internal contributions were the medical stop-loss sale (excellent timing), the two Fortitude cessions, the group-LTC enrollment freeze, and the buyback. The reversal now underway is also external, which is precisely why it is not controllable.

How stable are revenues?

Structurally stable, competitively fragile. Group benefits are yearly renewable term — every case is re-priced and can be re-bid annually. Persistency runs ~89–92% on the group book (~10-year average tenure) and ~74–77% on voluntary/worksite (~4 years). FACT. So roughly one case in ten leaves every year in group, and one in four at Colonial Life. Premium income grew 3.8% annually FY23→FY25, roughly nominal payroll growth. Stability comes from the arithmetic of a large in-force book, not from customer lock-in.

Outlook for products/services?

Mixed by line, and the mix is deteriorating. FACT: group LTD premium fell 2.2% over two years with sales −17.7% in FY2025, while STD (+12.5%), group life (+11.4%), individual disability (+16.7%) and dental & vision (+16.8%) grew. INTERPRETATION: the shift out of the longest-duration, highest-margin, most-defensible line into shorter-tail, lower-margin, more commoditized lines is structurally margin-dilutive — the group disability income ratio fell from 20.4% (FY23) to 13.2% (Q1-26).

How big will this market be — growing, shrinking, domestic or international?

US group disability in-force is ~$19.9bn growing ~4.7%; group term life ~$32.8bn growing ~1%; workplace supplemental health ~$3.3bn of new premium. FACT. LIMRA projects employment growth below 1% for 2026–2028 and life in-force premium growth “below historical averages,” and 2025 new-sales premium turned negative across life (−2%), disability (−5% through 9M) and supplemental health (−5%). FACT. ~91% of Unum’s premium is US; the UK is ~9% of core AOI in a market that grew 2.0% in 2025 with group income protection lives actually falling 1.6%. FACT. INTERPRETATION: this is a low-single-digit-growth, employment-linked market with no international growth optionality of consequence.

Business Quality & Competitive Moat

Is the industry getting more or less competitive?

More. No new entrants, but every scaled incumbent is expanding into the same pool at peak reported margin: MetLife’s “New Frontier” tilts explicitly to Group Benefits; Securian’s record 2025 was “anchored in group benefits”; Sun Life grew US group sales +27% to a record $1.8bn; Mutual of Omaha grew group premium +43% in two years (17% five-year CAGR); StanCorp/Meiji Yasuda bought scale three times in four years, including Allstate’s Employer Voluntary Benefits for $2.0bn. FACT. The price evidence is decisive: in 2024, LTD in-force premium-per-life rose +3.5% to $292.3 while new-sales premium-per-life fell −13.5% to $214.2 — new business written ~27% below the in-force rate. FACT. In Marathon terms: no new capital, but rising incumbent capacity commitment, already biting on price.

How profitable is the business (ROIC, ROE)?

For an insurer, ROIC is not meaningful; ROE is the right measure, and it must be looked at three ways.

FACT for the first four; INTERPRETATION for the fifth. The 20.5% headline excludes 43% of shareholder equity. The bridge from 6.7% to 20.5% is +6.0 points for excluding $667.5M of after-tax charges, −1.6 points for adding back AOCI, then +9.4 points for dropping the Closed Block’s and Corporate’s capital out of the denominator. Unum’s 12-year GAAP ROE average is 9.33%, and every sub-6% year (2014, 2018, 2025) is an LTC charge year. Against a ~9.5% cost of equity, this company has earned approximately its cost of capital across a full cycle and no more.

How profitable is the industry — how many competitors, what barriers to entry?

Group benefits runs 6–10% segment margins at 15–22% segment ROE. Hartford guides its long-run Employee Benefits net income margin down to 6–7% and printed 6.4% in Q1-26. Tokio Marine’s newly-disclosed IFRS “Employee Benefits” line runs a 95.3% combined ratio versus 85.4% for its Specialty P&C. FACT. Top-10 carriers hold ~75% of new disability premium and ~64% of workplace life new sales — moderately concentrated. Barriers are partial and weak: real infrastructure, data, licensing and broker-relationship barriers exist (hence zero new entrants), but they are barriers to entering at scale, not to competing for any individual case.

Can the business be easily understood?

The operating businesses, yes — premium in, claims out, annual repricing. The Closed Block cannot be understood from the outside, and that is the investment problem. It requires estimating morbidity, mortality, lapse, claim resolution rates and regulator-approved rate increases twenty years forward on 835,000 lives. Management discloses sensitivities but not the one number that matters (the capped-cohort share). The block runs to peak industry claims around 2041.

Can it be undermined by foreign low-cost labor?

No. Claims adjudication is US-regulated, licensed, and ERISA-governed. Not a risk vector.

Do brands matter?

No. The buyer is a corporate benefits manager advised by a broker, running a structured RFP. Unum’s own 10-K lists price among the principal competitive factors and states “Some of these companies have more competitive pricing.” FACT. Colonial Life’s worksite model, which sells to the employee, has marginally more brand relevance — and earns 18.2% allocated ROE against Unum US’s 22.6%, which argues against brand being where the value is.

What is the nature of competition?

Annual, broker-intermediated, price-led auction. New-sales rankings churn ±40–60% year over year even as in-force share barely moves — because in-force is a slow-moving average of many years’ pricing at ~90% persistency, while the bid is genuinely contested. FACT. And the value leaks structurally: the employer, not the insured, is the buyer, and the employer’s agent — the broker — is compensated on placement and sells the employer on extracting carrier concessions at renewal. The agency relationship that normally creates manufacturer pricing power runs the wrong way here. Broker distribution has meanwhile concentrated into roughly six balance sheets in thirty months (Aon/NFP $13.4bn, MMC/McGriff $7.75bn, AJG/AssuredPartners $13.45bn, Brown & Brown/Accession $9.825bn, OneDigital >$7bn, Hub at $29bn), concentrating placement leverage exactly as carriers most need rate. FACT.

Customers’ switching costs?

Real but modest, and demonstrably purchasable. ~90% group persistency implies genuine friction — transferring disabled-lives reserves, re-enrolling employees, re-integrating HRIS feeds. But the decisive evidence is management’s own: Christopher Pyne, explaining the Q1-26 persistency improvement, said “we can keep customers with a modest rate reduction going forward at very high margins,” while the Q1-26 10-Q attributes the worsening group disability benefit ratio to “pricing actions.” FACT. Retention bought with a rate cut is a discount, not captivity. Confirming the elasticity: when Unum pushed rate in FY2025, LTD persistency fell 93.3%→91.1%, STD 91.7%→88.9%, group life 92.0%→90.2%, and LTD sales fell 17.7%. FACT. Note the asymmetry by case size: MetLife reports high-90s persistency and 20-year average tenure in national accounts — switching costs rise sharply with case size, and Unum’s blended ~90% implies the mid-market is where the churn lives.

Greenwald verdict: supply/cost advantage in disability claims management is real and partial (Unum US 12.6% after-tax segment margin vs Hartford’s 8.4%). Customer captivity FAILS. Therefore the scale-plus-captivity conjunction fails, and scale cannot compound. Network effects: none. Brand: negligible. Government protection (ERISA preemption): present, real, and non-differentiating — it lifts the whole industry’s profit pool. Conclusion: operational effectiveness, which is emulable, not a moat.

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet?

Two, both soft. (i) The embedded assumed-future-rate-increase asset inside LTC reserves — management’s ±10% sensitivity of $140M implies ~$1.4bn if roughly linear (INTERPRETATION), of which only ~15% of the program has been achieved. This is an asset contingent on fifty state regulators. (ii) Fairwind’s Vermont permitted-practice benefit of ~$185M, which inflates the captive’s statutory surplus and does not exist under NAIC statutory accounting (down from $334M in 2024). FACT. That one arguably belongs in the liability column.

Off-balance-sheet liabilities?

• $643.8M of FHLB advances excluded from management’s leverage ratio — including them takes debt/capital from 25.3% to 25.4% (GAAP). FACT. Small but a presentation choice.
• $756.7M of unfunded private-equity commitments. FACT.
• A $125M NPV-capped experience-volatility cover provided to the Fortitude II retrocessionaire, settled every five years and trust-secured, for a $5M fee. FACT. Risk was not fully transferred.
• Reinsurance recoverable concentration: ~72% in essentially two counterparties. Mitigated by Fortitude’s A rating, a collateralized trust for Unum’s benefit, and 100% retrocession of biometric risk — but the identity of the ultimate risk-taker is undisclosed on both deals.

How conservative is the accounting?

Mixed, and the pattern is directional. The conservative-looking headline (five straight years of core-segment reserve releases) sits alongside three LTC charges in four reviews. Pre-tax, over three years: Unum US +$420.1M, Colonial Life +$135.6M, Closed Block LTC −$837.1M. FACT. INTERPRETATION: five years of over-reserving in the good businesses have been funding five years of under-reserving in the run-off block — and management excludes both legs from its headline metric, so neither is visible.

Three further presentation choices deserve flagging, all FACT:

1. Reserve assumption updates are excluded from adjusted operating income despite appearing in five of five years and averaging −$99.4M pre-tax annually. A cost that recurs annually is an operating expense.
2. Amortization of the cost of reinsurance ($116.7M pre-tax, $0.53/share, and escalating) is excluded — removing the price tag of a strategy while keeping its benefit.
3. In Q1-2026 management redefined adjusted operating income to exclude the entire Closed Block segment — in the exact quarter that segment swung from +$8.0M to −$145.3M. Prior periods were restated, cutting the FY2025 base from $8.13 to $7.93. The block still carries $16.7bn of LTC reserves and consumes real statutory capital.

The asset side, by contrast, is genuinely conservative: below-investment-grade at 2.8%, a $5.9M credit allowance on a $34.7bn bond book, office CRE at 0.7% of total investments. FACT.

How CapEx-hungry is the business?

Not applicable in the manufacturing sense — depreciation runs ~$123M/yr on a $13.1bn revenue base. The correct sector analog is statutory capital intensity, and here the answer is severe and specific: the Closed Block consumes $4,830.0M of allocated equity — 38% of the group total — to earn $40.7M, a 0.8% return. FACT. That is the capital hunger that matters. Technology spend is real but modest (HR Connect, claims platforms) and is running through the expense ratio, which has been flat at 22.5%/22.5%/22.6% for three years.

Capital Allocation & Management

How much FCF does the business generate, how does management use it, what is the philosophy?

For an insurer, “free cash flow” is subsidiary dividends to the holding company, not GAAP operating cash flow — and GAAP OCF ($687.7M in FY2025) is unusable here anyway, distorted by Fortitude Re moving $935.5M of cash out. FACT.

Unum Group received $1,547.1M in subsidiary dividends in 2025, but $787.9M was extraordinary (including $630.0M from First Unum, approved only after First Unum ceded 100% of its LTC book to Provident). It returned $1,311.4M ($1,011.7M of buyback at an average $74.39, plus $306.2M of dividends) against ~$555.4M of ordinary FCF — 2.36x. FACT.

The forward constraint is the important number: 2026 ordinary dividend capacity is $631.4M, down 54% (a formula consequence of the 52% statutory earnings drop), plus ~£125M — roughly $800M — against ~$1.3bn of guided return. FACT. INTERPRETATION: the buyback is running ~$500M/yr ahead of flow and is partly funded from a stock of capital — $2,344.1M of holdco liquidity guided down to $2.0–2.5bn, and RBC guided from 440% down to 400–425% with no committed timeline.

Philosophy: return ~100% of free cash flow, second consecutive year; open-ended $1.0bn authorization (the first without an expiry); dividend raised ~10% to $0.505/quarter ($2.02, ~2.3% yield), with an 8.4% DPS CAGR since 2019.

Significant acquisitions recently?

Essentially none for a decade, and this is the strongest single item in the capital-allocation case. One $145.4M outflow in FY2018 (Unum Poland; $12.6M of premium); Starmount (2016) immaterial; no acquisition line FY2019–FY2025. FACT. In an industry with at best narrow barriers, where peers have been buying scale aggressively — Meiji Yasuda/StanCorp three times in four years, Sun Life’s $2.6bn DentaQuest purchase (since impaired by $61M and a collapsed earnings target), Voya’s Benefitfocus (described by Bloomberg as “ill-fated”) — not empire-building is the correct Greenwald answer and management has held that line.

The countervailing transactions are risk-transfer, not acquisition: Fortitude I ($3.4bn statutory, ~17% premium, $848.2M cost of reinsurance) and Fortitude II ($3.8bn statutory / ~$4.5bn best-estimate for ~$5.7bn of assets, ~27% premium). INTERPRETATION: the price got worse, not better, on the second deal despite eighteen additional months of rate approvals — evidence Unum is a price-taker in a market one-to-two balance sheets deep.

One new item to flag but not yet weight: a $500M funding-agreement-backed note program established February 2026 for “spread lending” — levered investment income with no competitive advantage. Nothing drawn. FACT.

Buying back shares?

Aggressively, and the record is genuinely excellent — but its conditions have expired.

FACT. 60.3M shares — 19% of the company — retired at a blended $54.66 against today’s $89.05: roughly $2.07bn of value created, with nothing bought in the 2020 panic. But 72.6% of all dollars were deployed in 2024–Q1’26 at an average $67.94 versus $34.09 for 2019–23, and the multiple paid has climbed monotonically from 0.45x to 0.96x adjusted book.

INTERPRETATION, and it matters: at ROE ≈ COE, a buyback cannot create value — it can only transfer it, adversely so above intrinsic value. The justified P/B is nearly insensitive to growth (0.92x/0.94x/0.93x/0.91x at g = 2/3/4/5%). Historic buyback skill is not a permanent attribute; it came from buying a de-rated block, and the de-rating is gone. Asked why he was buying at $89, McKenney called it “attractive prices” while offering no multiple, no book reference and no intrinsic-value frame — and when pressed gave a capital-availability rationale (“we sit in excess capital”). That reasoning justifies buying at any level. The Q2-26 10-Q monthly repurchase table (~2026-07-29) settles whether the program has made its first materially expensive purchase.

Issuing large amounts of new shares to insiders?

No. Share count fell from 202.9M (FY2019) to 160.75M (Q1-26). Aggregate insider ownership is under 1% (19 persons, 1,608,925 shares; the CEO 0.52%). Grants are routine annual equity awards. FACT.

Compensation policy of directors/management?

This is the first-order governance finding of the report. From the DEF 14A filed 2026-04-09:

Annual incentive: After-Tax Adjusted Operating EPS 50%, core Premium Income 20%, Sales 10%, Customer Experience 10%, Adjusted Other Operating Expense Ratio 10%. Standing exclusions, quoted verbatim: “The impact of any unplanned acquisitions, divestitures or block reinsurance transactions”; “Unplanned adjustments to the Closed Block”; “Unplanned reserve assumption updates”; “…or stock repurchase or issuance.” The long-term plan applies the same scrub to “Adjusted Book Value.” FACT.

The wedge: GAAP diluted EPS fell 54.9% to $4.27. The metric management is paid on was $8.13 — 90% higher. FY2025 payout was 76.4% of target. The 2023 long-term cycle paid 172.6% (adjusted-book ratio 129.0%, business results 143.8%, relative TSR at the 100th percentile), earning the CEO $7,767,000 in cash — over a 2023–25 window in which the LTC assumption line ran a cumulative −$837.1M. FACT.

FY2025 NEO awards (all at 76.4%): McKenney $2,154,995; Zabel $890,905; Pyne $620,750; Anderson $307,216; Iglesias $476,912. CEO SCT total $19,417,476; pay ratio 1-to-222. For 2026 the CEO’s AI target rose 250%→260% of salary and LTI from $11.3M to $11.651M. FACT.

The counterweights are real and should not be suppressed: the 76.4% is a genuine haircut; the 172.6% was market-validated by a 100th-percentile relative TSR against a named nine-company peer group; excluding buyback effects from an EPS metric is shareholder-friendly, not the reverse; 2026 adds PSUs with new targets. Hygiene is mostly good — McKenney holds 67.0x salary ($76.0M) against a 6x requirement, single share class, no pledging, anti-hedging, two clawback policies (Rule 10D-1 plus a broader misconduct policy requiring no restatement). Say-on-pay support was 96.1%, up from 92% — support rose in the year earnings halved.

The residual objection survives all of it: there is no GAAP, statutory, ROE, or reserve-adequacy measure anywhere in the plan. Relative TSR is a ±20% modifier on a scrubbed base, not a gate. INTERPRETATION: if the metric excludes the reserve charges, the plan design creates an incentive to under-reserve and to classify strengthening as non-operating. That is a structural criticism of plan architecture, not an allegation about any individual’s conduct.

Motivations of management?

Richard McKenney has been CEO since 2015-05-21 and holds $76.0M of stock — 67x salary — so his economic alignment with the share price is genuine and large. FACT. Full-tenure TSR is +261.4% / +12.20% annualized against the S&P 500’s +319.6% / +13.72% — underperforming the index by 1.51 points annually while ranking 3rd of 9 against a weak peer group (median +10.14%, containing Lincoln at +0.72%). The path is the story: +76.6% to January 2018, then −81.1% to March 2020 on the LTC charge plus COVID — the worst drawdown in the stock’s forty-year history, entirely within this tenure — then +979.8%. Ex-AOCI BVPS compounded 9.1%. FACT.

There is no CEO or CFO transition — zero Item 5.02 8-K filings after 2024-01-04. The 2026 Form 3s map to a Colonial Life segment president change (Steve Jones succeeding the retiring Tim Arnold), a new director (Kristi Matus), and an internal promotion (Andrew Walker). FACT.

Insider transactions: parsed from 112 raw ownership XML filings (avoiding the rendered-HTML trap flagged in prior engagements), 219 itemized lines. Zero code-P open-market purchases in thirty months. Fifteen sales across 2025–26 totaling $9,791,103, fourteen of fifteen discretionary; the largest was McKenney selling 50,000 shares in three tranches on a single day (2025-03-03) for $4,099,453. CFO Zabel sold 7,500 shares at $76.38 three weeks after the Q3 charge. FACT. The “selling into the high” narrative does not hold and should not be asserted — every sale executed between $72.47 and $82.75, all below today’s $89.05. Insiders sold on the way down and have been absent from the rally. The fairer indictment is the other side: at $72–73 in February–March 2026, with the charge fully public and the stock ~20% below spot, not one insider bought. At $9.8M against a $14.3bn market cap this is routine diversification; the signal is composition, not magnitude, and should not be over-weighted.

Valuation & Market Data

Is the stock an ADR, MLP, or K-1 issuer?

No. Unum Group is a US-domiciled Delaware corporation, NYSE-listed, single share class, ordinary 1099 dividend treatment. Not an ADR, not an MLP, no K-1. FACT.

Dividend policy?

Raised ~10% to $0.505/quarter ($2.02 annualized) effective Q3-2026, a ~2.3% yield. DPS has compounded 8.4% since 2019 ($1.09 → $2.02) while the aggregate cash cost rose only 34%, because the share count fell 19%. FACT. FY2025 payout was 41.5% of depressed GAAP EPS (16.7% in FY2024), or ~25% of adjusted. The dividend is comfortably covered even on normalized earnings; it is the buyback, not the dividend, that is outrunning cash generation.

How profitable is the business?

Covered above. In one line: a 20.5% ROE operating franchise inside an 11.1% ROE (reported) / ~9.1% (normalized) company, because 38% of the equity earns 0.8%.

Is net income diverging from cash from operations?

Yes, and in a way that requires the right lens. GAAP net income of $738.5M against GAAP OCF of $687.7M looks close — but both are distorted. FY2025 OCF fell from $1,513.2M principally because Fortitude Re took $935.5M of cash out of the company; net income fell principally because of a $377.8M after-tax non-cash reserve charge. FACT. Neither figure is decision-useful.

The correct sector metric — subsidiary dividends to the holding company — tells the real story: $1,547.1M received in 2025, of which $787.9M was extraordinary and regulator-approved, against $1,311.4M returned. FACT. And the forward figure is the constraint: 2026 ordinary capacity of $631.4M, down 54%. INTERPRETATION: the divergence that matters is not net income versus OCF; it is capital return versus statutory dividend capacity, and that gap is ~$500M a year.

Risks & Downside

What factors would cause the stock to decline?

Ranked by expected impact:

1. The LTC net premium ratio reaching the 100% cap (97.6% now, ~1.6pp/yr drift, ~2.4pp headroom ≈ ~18 months), converting the block into an instrument where adverse deviation hits earnings immediately and favorable deviation cannot be banked. The Q1-26 −$145.3M Closed Block loss was the first observation of that regime.
2. A fourth LTC reserve charge in five years at the Q3-2026 annual review. Three of four recent reviews were charges; industry peers (Genworth, Manulife) are still strengthening, and Genworth’s Q4-25 miss was substantially experience variance rather than assumption change — meaning current experience is running worse than already-strengthened assumptions.
3. Confirmation that group LTC (~$7.7bn) has no buyer. No group-LTC transaction exists anywhere in the 2023–26 industry set.
4. Group disability incidence rising with the labor cycle. STD incidence is already rising and leads LTD by the 90–180-day elimination period by construction; unemployment is ~4.9% with employment growth forecast below 1%.
5. A buyback cut, forced by the $631.4M dividend-capacity constraint.
6. A larger-than-expected GAAP charge on the Fortitude II closing.
7. Factor-regime mean reversion — Value and DividendYield, Unum’s two largest style loadings, are both at ~+1.7 standard deviations of favor.

Risk of a catastrophic loss?

Low in the solvency sense, high in the multi-year-earnings sense — and the distinction matters. Solvency risk is genuinely low: RBC 440%, holdco liquidity $2,344.1M, no debt maturity before 2028, all four rating agencies stable (senior bbb+/BBB/Baa2/BBB; financial strength A/A/A2/A), below-investment-grade at 2.8%, office CRE 0.7% of investments. FACT.

The tail risk is not insolvency; it is a multi-year sequence of reserve strengthenings on a block running to peak industry claims around 2041. That is precisely what produced the −81.1% drawdown of 2018–2020 — the worst in the stock’s forty-year history — with comparable episodes in 2002–03 and 2008–09. This has happened twice before to this company on this same liability. Management’s own stacked sensitivities (lapse+mortality ±$430M, incidence ±$370M, resolutions ±$290M, unapproved rate increases ±$140M) sum to roughly $1.2bn of pre-tax exposure before touching the implied ~$1.4bn rate-increase asset.

Chance of a total loss?

Negligible. This is a 178-year-old, investment-grade, well-capitalized insurer with a clean asset portfolio, stable ratings, positive statutory earnings, and no near-term maturities. The realistic bear case is a return to the 0.6–0.9x book multiple at which the stock traded as recently as FY2023 — which is a large drawdown, not an impairment of the enterprise.

Recent News & Events

Has the business environment changed recently?

Yes, in three material ways, none of them favorable and none fully in the price.

1. The margin cycle turned. All three tailwinds (group life mortality, disability incidence/recovery, discount rates) peaked in FY2024 and are reversing. Benefit ratios are deteriorating in group disability (59.0%→63.7%), international (69.0%→73.5%), supplemental & voluntary (39.8%→49.9%) and dental & vision (73.1%→75.6%) — four lines across two geographies simultaneously. Independently corroborated at Hartford (disability LR 67.1%→72.7%), Lincoln (70.1%→73.4%) and MetLife (IABR 72.2%→75.8%). FACT.
2. The LTC block passed an inflection. NPR to 97.6%, capped cohorts now costing $436.7M pre-tax and appearing in Colonial Life as well, and the Q1-26 Closed Block loss of $145.3M. FACT.
3. Capital capacity halved. 2026 ordinary dividend capacity fell 54% to $631.4M; FY2025 statutory earnings missed guidance by 15–30% ($1.1bn against a $1.3–1.6bn guide). FACT.

Adjacent and instructive: medical stop-loss blew up industry-wide to a record ~91% loss ratio in 2025, Gallagher Re reported three major reinsurers signaling exit from US medical reinsurance in February 2026, Voya’s stop-loss unit is under activist pressure to be sold, and Sun Life declined to disclose its FY2025 loss ratio. Unum sold its stop-loss business to Amynta in July 2024, immediately before all of this. FACT — and the single best evidence of management quality in this file.

Significant acquisitions?

None. Two risk-transfer transactions instead: Fortitude I ($3.4bn statutory LTC + ~$120M IDI premium; closed July 2025; $848.2M cost of reinsurance) and Fortitude II ($3.8bn statutory / ~$4.5bn best-estimate; announced 2026-07-02; closing during 2026; GAAP charge undisclosed). Cumulatively ~40% of LTC statutory reserves ceded in eighteen months. The stock fell 3.1% on the second announcement. FACT.

Change in accounting policies?

Yes — and it is the most consequential disclosure event of the period. In Q1-2026 management redefined adjusted operating income to exclude the entire Closed Block segment, and simultaneously stopped adjusting for cost-of-reinsurance amortization “because the majority of these items are included in Closed Block results.” Prior periods were restated, cutting the FY2025 base from $8.13 to $7.93. The change was made in the exact quarter the Closed Block swung from +$8.0M to −$145.3M. The volatile alternatives portfolio was moved below the line in the same reset. FACT.

INTERPRETATION: the FY2026 guide of $8.60–8.90 represents +8–12% growth off the redefined $7.93 base but only +5.8% to +9.5% off the as-reported $8.13. An analyst on the call described the comparison as “tough to compare.” Post-2026 series will not be comparable to what came before, and any aggregator’s TTM “adjusted” figure for UNM is unreliable, because the basis change was never applied to FY2025 in any published document.

Recent changes — new markets, facilities, management?

• Group LTC closed to new enrollments on existing cases effective 2026-02-01 — a deliberate tail-risk reduction that costs premium. Q1-26 saw 7% of cases terminate, ~30,000 net lives, with a statutory reserve release of “less than $100 million.” FACT.
• New $1.0bn buyback authorization (announced 2025-12-04, effective 2026-01-01) — the first without an expiry date. FACT.
• $300M of 5.250% senior notes due 2035 issued November 2025 (refinancing; coupon stepped up 137bp from 3.875%). AM Best affirmed “bbb+”/stable. No adverse ratings action from any of the four agencies despite the charge. FACT.
• Dividend raised ~10% to $0.505/quarter (2026-05-21). FACT.
• $500M funding-agreement-backed note program established February 2026 for “spread lending,” nothing drawn — the first sign of reaching outside the franchise for earnings. FACT. Flag; do not yet weight.
• Management: no CEO or CFO change. Segment-level only — Steve Jones to President of Colonial Life (2026-06-08) as Tim Arnold retires after 41 years; Andrew Walker promoted to EVP Chief Customer Operations Officer; Kristi Matus elected to the board. FACT.

One timing note that qualifies everything above: Unum reports Q2-2026 on 2026-07-28 with the call on 7/29 — ten days after this article date. This analysis is written into a blackout. The Q2 10-Q’s monthly repurchase table and the Q3-2026 assumption review are the two near-term artifacts most likely to move the thesis.

APPENDIX B — Source Appendix

Unum Group (NYSE: UNM) · 2026-07-18

All public sources consulted in the preparation of this article. Primary sources are listed first within each category. Every non-obvious factual claim in the analysis and in Appendix A traces to an entry below.

1. Unum Group primary filings (SEC EDGAR, CIK 0000005513)

Full trailing 60-month EDGAR corpus reviewed: 5 × 10-K, 15 × 10-Q, 52 × 8-K, 5 × DEF 14A, and 112 Form 3/4/5 ownership filings parsed from raw XML (219 itemized transaction lines). Per the standing methodology note, SEC Form 4s mirrored as .xml are rendered HTML; all insider figures in this article were derived from the raw ownership documents and reconciled to totals.

2. Unum Group company communications

3. Industry data and surveys

4. Peer and competitor primary filings

The Hartford (HIG, CIK 0000874766) — FY2025 10-K (2026-02-20); FY2025 Investor Financial Supplement (2026-01-29) https://www.sec.gov/Archives/edgar/data/874766/000087476626000006/ex992ifs12312025.htm; Q1-2026 IFS (2026-04-23); FY2024 IFS (2025-01-30); Q4/FY2025 and Q1-2026 earnings releases; Q4-2025 and Q1-2026 call transcripts.

MetLife (MET, CIK 0001099219) — Q1-2026 Quarterly Financial Supplement (2026-05-06) https://www.sec.gov/Archives/edgar/data/1099219/000109921926000032/ex992qfsq126.htm; Q4/FY2025 QFS (2026-02-04); Q4/FY2024 QFS (2025-02-05); Q4-2025 and Q1-2026 call transcripts.

Lincoln Financial (LNC, CIK 0000059558) — Q4-2025 earnings release and statistical supplement (2026-02-12) https://www.sec.gov/Archives/edgar/data/59558/000005955826000007/a4q2025lncearningspr.htm; Q1-2026 release and supplement (2026-05-07); Q4-2024 release and supplement (2025-02-06); FY2025 10-K; Q4-2025 and Q1-2026 call transcripts.

Principal Financial (PFG, CIK 0001126328) — FY2025 10-K (2026-02-18) https://www.sec.gov/Archives/edgar/data/1126328/000110465926017031/pfg-20251231x10k.htm; Q4-2025 release and 2026 outlook (2026-02-09); Q1-2026 release (2026-04-23); Q4-2025 call transcript.

Aflac (AFL, CIK 0000004977) — Q1-2026 Financial Analysts Briefing supplement (2026-04-29) https://www.sec.gov/Archives/edgar/data/4977/000162828026028396/afl033126-fabdocument.htm; Q1-2026 and Q4-2025 CFO video-update exhibits; Q1-2026 call transcript.

Voya Financial (VOYA, CIK 0001535929) — 8-K and stop-loss supplemental slides (2024-12-09) https://www.sec.gov/Archives/edgar/data/1535929/000153592924000115/unm-20241206.htm and /exhibit991.htm; Q1-2026 Investor Supplement and press release (2026-05-05); Q4-2025 press release (2026-02-03); FY2025 10-K (2026-02-20); 8-K (2026-07-08); Q4-2024, Q2-2025, Q4-2025 and Q1-2026 call transcripts.

Sun Life Financial (SLF, CIK 0001097362) — 2025 Annual MD&A on Form 40-F (2026-02) https://www.sec.gov/Archives/edgar/data/1097362/000109736226000010/a2025q4slfmdalive.htm; Q4-2025 earnings release 6-K; Q1-2026 earnings release 6-K; Q3-2025 and Q2-2025 shareholders’ reports; Q4-2024, Q2-2025, Q3-2025, Q4-2025 and Q1-2026 call transcripts.

CNO Financial (CNO, CIK 0001224608) — Form 10-Q, period ended 2025-09-30 https://www.sec.gov/Archives/edgar/data/1224608/000122460825000070/cno-20250930.htm (Bermuda reinsurance transaction; Worksite fee-services exit).

Genworth Financial (GNW, CIK 0001276520) — FY2025 10-K https://www.sec.gov/Archives/edgar/data/1276520/000162828026012828/gnw-20251231.htm (LTC reserve strengthening; Enact stake as SOTP validation).

Manulife (MFC, CIK 0001086888) — FY2024 Form 40-F https://www.sec.gov/Archives/edgar/data/1086888/000108688825000054/mfc-20241231.htm; Global Atlantic and RGA LTC block transaction releases.

Marsh McLennan (MMC, CIK 0000062709) — FY2024 10-K (McGriff acquisition). Arthur J. Gallagher (AJG, CIK 0000354190) — 8-K, AssuredPartners closing (2025-08-18). Brown & Brown (BRO, CIK 0000079282) — 8-K, Accession/RSC Topco closing (2025-08-01). Ryan Specialty (CIK 0001849253) — FY2025 10-K (medical stop-loss MGU platform).

Foreign-parent disclosures: Tokio Marine Holdings FY2024 and FY2025 results presentations (Delphi Financial Group segment; new IFRS “North America — Employee Benefits” line); Reliance Standard Life FABN investor updates as of 6/30/2025 and 12/31/2025; Meiji Yasuda Life Financial Results Summaries for FY ended 2025-03-31 and 2026-03-31 (StanCorp reference pages); Meiji Yasuda Annual Report 2025; Meiji Yasuda/StanCorp–Allstate EVB acquisition release (2024-08-14).

5. Quantitative data sources

Methodological notes on the quantitative sources, recorded for reproducibility:

• ROIC.ai’s generic template fields (gross_margin 100%, EBITDA, is_nonop_income_loss) are meaningless for an insurer and were not used. Its return_on_cap is not a meaningful ROIC for a life insurer. Its ROE series does not reconcile to net income over average or ending GAAP equity on any basis and was rejected in favor of figures computed directly from the filings. Its reported revenue for FY2025 ($12,984.1M) differs from the 10-K ($13,075.5M) by $91.4M, unexplained; the 10-K governs. Its bs_mkt_sec_other_st_invest of $36,072.8M matches no 10-K subtotal (actual total investments $43,520.8M). Its EPS, AOCI, equity and share-count fields do tie and were used.
• ROIC.ai’s enterprise-value tool returns −$19.1bn for UNM and is unusable for an insurer; the sum-of-the-parts was therefore built on an equity basis.
• Peer P/B figures were recomputed from raw common equity ÷ actual shares outstanding, because the aggregator’s own book-value-per-share and price-to-book fields use share counts that do not tie (UNM’s implied 136M against 160.75M actual).
• FactorsToday leaderboard returns and Sharpe ratios are annualized at every horizon including m3; the 3-month figure was de-annualized (+60.1% annualized = +12.5% actual quarter) and reconciled against the AZI price series before use.
• The AZI P/E percentile is distorted by charge-depressed TTM EPS ($4.60) and was explicitly discounted in favor of the P/B reading; the P/B percentile is the clean signal.

6. Trade press and secondary sources

Insurance Business (stop-loss market shifts, 2025-09-12; US medical reinsurance withdrawals, 2026-02-24; Securian 2025 results, 2026-03-12; New York Life record earnings, 2026-03-19); BenefitsPRO (Sun Life stop-loss repricing, 2026-02-13; Voya stop-loss unit sale speculation, 2026-05-07; Voya 2026 rate increases); Insurance Journal (TOMS Capital letter to the Voya board, 2026-06-01; Marsh/McGriff, 2024-11-18; AJG/AssuredPartners, 2025-08-18); Semafor (Voya takeover interest, 2026-07-16); Bloomberg via Insurance Journal (TOMS Capital campaign); S&P Global Market Intelligence (Manulife LTC deal rarity, 2025-01; US life insurance 2026 outlook, 2026-01); Reinsurance News (Sands Point/BRM; Meiji Yasuda/Banner Life); Healthcare Brew (stop-loss cost crisis, 2025-05-21); BNN Bloomberg (Sun Life Q4-2024 reaction, 2025-02-13); Investing.com and Motley Fool (earnings-call transcripts for HIG, LNC, PFG, SLF, VOYA, AFL); PR Newswire and BusinessWire (transaction announcements throughout).

7. Explicit limitations of this research

Recorded so that the reader can weigh the analysis appropriately:

1. Q2-2026 has not been reported. Unum reports on 2026-07-28 (call 7/29), ten days after this article date. This analysis is written into a blackout. Peers are likewise unreported: PFG 7/27, LNC 7/30, VOYA 8/4, AFL 8/6, SLF 8/6.
2. The share of LTC cohorts already at the 100% net-premium-ratio cap is not disclosed and could not be derived. It is the single most important gap in the file.
3. The GAAP loss on the Fortitude II closing is not disclosed. The ~$1bn estimate in the memo is explicitly labeled an inference from the asset-liability gap and the Fortitude I precedent.
4. The Milliman group-disability survey has a moving panel. The 2025 edition (2024 data) dropped The Hartford, Sun Life and Reliance Matrix. Unum’s apparent 2024 #1 ranking is therefore substantially a disclosure artifact; the memo states the honest conclusion (#1 or #2, too close to call) rather than the survey headline. The 2019–2023 share series was reconstructed on a consistent five-carrier panel.
5. Group life market share by premium at two points in time could not be obtained. AM Best ranks by face amount in force or issued — a poor economic proxy — and its full tables are behind bot protection. The Greenwald share-stability test for group life is not satisfiable from public sources, and no group-life share trend is asserted anywhere in this article.
6. LIMRA carrier-level league tables are paywalled. Only aggregate figures are public.
7. Portfolio duration and the duration gap are not disclosed by Unum; the inference that the LTC block is under-matched at the long end rests on the observation that 68% of derivative notional hedges LTC reinvestment risk.
8. “Average allocated equity” is a management construct, not an audited balance-sheet figure. The 22.6% / 18.2% / 14.8% segment ROEs and the 20.5% core figure are therefore soft; the total-company 11.1% is the reliable number because its denominator ties to the audited balance sheet. The sum-of-the-parts leads with the P/E lens for this reason.
9. The sum-of-the-parts has one degree of freedom. Core multiple and implied Closed Block value are perfectly substitutable; the analysis establishes a joint constraint, not a point estimate.
10. Names of the three reinsurers exiting US medical reinsurance were not obtainable. Gallagher Re does not name them; only RGA is identifiable and only via a secondary source that could not be fully loaded. Claims that Zurich or Swiss Re are among them appeared only in search paraphrase with no retrievable source and are not asserted in this article.
11. A minor unreconciled item: the FY2025 10-K states Fortitude I cash transferred as both $953.5M and $935.5M in different sections.
12. Any aggregator’s TTM “adjusted” EPS figure for UNM is unreliable, because management’s Q1-2026 basis change (excluding the Closed Block) was never applied to FY2025 in any published document. TTM GAAP net income is $781.4M; a clean TTM adjusted figure cannot be computed.