Aflac Inc AFL

Thesis

Aflac is two businesses stapled together: Aflac Japan, the dominant supplemental cancer/medical insurer in Japan with ~70% of operating earnings, and Aflac US, a payroll-deduction supplemental insurer sold through small-employer worksites. Both share the same DNA: small-ticket, indemnity-style policies that pay cash directly to policyholders when defined health events occur. Premiums are tiny ($30-100/month), persistency is high, distribution is locked into payroll deduction or Japan Post / Dai-ichi banking channels, and the underlying insurance risk is benign and well-modeled (cancer incidence in Japan, accident frequency in the US). The float is large and long-tailed, the loss ratio is stable, and capital requirements grow modestly.

The scorecard tells the story. ROIC 10y average is 10.18%, P/E TTM is 11.72 vs. 10y average 9.72, and the reverse-DCF implies -2.88% growth — the market is pricing AFL as if it shrinks in perpetuity. Owner earnings TTM are $5.44B against a market cap near $60B. Composite is 68/100 with strong valuation (20/30) and balance-sheet (18/25) sub-scores. The IV range — low $147, base $261, high $392 — implies a P/IV of 0.43, which is the cheapest tier of the screen. Even taking the low-case IV at face value, the stock offers ~30% upside plus a ~2% dividend and a buyback that has shrunk the share count meaningfully (despite the noisy 10y change of +3.52% which reflects share-issuance from old SOP plans).

The price/IV math: at $112.88 vs. base IV $260.98, you pay 43 cents for a dollar of conservatively-modeled value. Even haircutting the base case 40% for yen translation risk gives ~$157 — still 39% above today. This is a buy with sizing constrained by FX, not by franchise quality.

Moat

Aflac's moat is real but structurally different from the auto-insurance moats Buffett describes for GEICO. Buffett's framework for insurance excellence is the four-part discipline: understand exposures, evaluate likelihood conservatively, price for profit, and walk away when premiums are inadequate [2]. Aflac passes all four in its core supplemental product because the underlying insured event — diagnosis of cancer or hospitalization — is well-characterized actuarially and the premiums are small enough that price competition rarely forces irrational quoting.

Cost advantages (NARROW-to-WIDE in Japan). In Japan, Aflac is the dominant cancer insurer with roughly 70%+ market share in the cancer-policy segment, sold through Japan Post Holdings (an exclusive alliance covering ~20,000 post offices) and through Dai-ichi Life. This distribution is functionally irreplaceable for a new entrant — you cannot just call Japan Post and ask to share the counter. The economics resemble GEICO's in that scale lowers per-policy expense, but unlike GEICO the channel itself is exclusive rather than direct-to-consumer [1]. New entrants would need to build an equivalent agency network or buy distribution; neither is plausible at the small premium amounts involved.

Switching costs (NARROW). US worksite distribution creates real, if not enormous, switching costs. Once Aflac is enrolled in a small employer's payroll-deduction system, the HR friction to switch — re-enroll every employee, re-train benefits administrators, re-paper policies — is meaningful. Aflac claims roughly 50,000+ payroll accounts. Persistency runs in the 75-80% range. Not a network effect, but a real per-customer stickiness similar to small-business software vendors.

Intangibles (NARROW, brand-only). The Aflac duck is one of the most-recognized brand mascots in US advertising, comparable in awareness — though not category power — to GEICO's gecko [1][6]. Brand awareness compresses CAC at the worksite level because employees recognize the product before the agent walks in. In Japan the brand premium is real among older policyholders but eroding among younger demographics who buy through banks and digital channels.

Pricing power (NARROW). Premium is small in absolute terms ($30-100/month), so customers are not price-shopping the way they shop auto insurance. Aflac has historically been able to push through modest annual premium increases without dramatic lapse. However, the yen-denominated Japan premium stream cannot be raised aggressively because it competes with lower-margin medical riders and because the Japanese regulator scrutinizes pricing.

Network effects (NONE). No two-sided marketplace dynamic; supplemental insurance does not get more valuable as more people buy it.

Competitor stress test ($10B + 5 years). Could a well-capitalized entrant — say, MetLife with $10B and five years — replicate Aflac Japan? No. The Japan Post relationship dates to 2008, was deepened in 2013 with cross-shareholding, and now spans ~20,000 outlets. Building a competing distribution network from scratch in Japan would face entrenched incumbents (Dai-ichi, Nippon Life), regulatory friction for a foreign entrant, and the deflationary economics of small-ticket supplemental in a flat-population market. Could a US entrant attack Aflac's worksite franchise? Easier — Allstate, Unum, Colonial Life all play here — but Aflac's brand and account density create a four-to-five-year lag for any rational attacker, by which time Aflac has rolled out new product riders.

Erosion risk. The structural risk is demographic and channel. Japan's population is shrinking 0.5%/year and aging; new policy growth is hard. Younger Japanese consumers buy supplemental coverage via online comparison sites, where Aflac's distribution moat is narrower. In the US, the worksite distribution model is being squeezed by HR-tech aggregators (Gusto, Rippling) that may eventually disintermediate the agent.

Applying Buffett's test on insurance discipline [2][3], Aflac has historically walked away from underpriced business in Japan — refusing to chase low-premium medical riders during the 2010s — which is the discipline he says most insurers flunk [2]. That history, combined with a structurally exclusive distribution franchise and rational underwriting, supports a real but bounded moat.

Moat verdict: NARROW

Management & Capital Allocation

Aflac's capital allocation under CEO Daniel Amos (CEO since 1990, Chairman since 2001 — 35-year tenure) and CFO Max Broden has been disciplined-by-insurance-standards but mediocre-by-best-in-class compounder standards. The five-choice framework:

1. Reinvestment in the operating business. Modest. The supplemental insurance business is not capital-intensive at the underwriting level; growth requires growing in-force premium, not building factories. Reinvested capital primarily funds technology (SmartClaim, online quoting), regulatory capital cushion in Japan, and product development. Returns on reinvested incremental capital are bounded by market growth — Japan supplemental is structurally flat — so management has correctly avoided force-feeding capital back into a saturated business.

2. Acquisitions. Largely absent, which is good. Aflac has not pursued large M&A. The Trupanion-style temptation to roll up adjacent insurance lines has been resisted. The largest strategic moves are the Japan Post alliance (2008, deepened 2013 — a partnership, not an acquisition) and the Zurich Japan acquisition (2017, $945M, a small bolt-on for cancer policies). Discipline grade: B+.

3. Debt. Aflac runs with modest leverage by P&C-insurer standards — debt-to-capital around 20% — and net debt to EBITDA shows -156x, which simply means net cash position dominates. The company holds ~$130B+ of investments against ~$8B of debt. Interest coverage is reported as 0.0 in the scorecard, which is a screening artifact (insurance-company income statements don't map to standard EBITDA/interest formulas) rather than a stress signal. Balance sheet earned 18/25 — strong.

4. Buybacks. This is where the qualitative story is interesting. Aflac has been a persistent buyback compounder for two decades. Annual buyback authorizations have run $2-3B against a $60B market cap, implying a 3-5%/year share-count reduction in normal years. The 10y share-count change of +3.52% looks like it contradicts this, but that figure is gross-of-issuance and includes legacy stock-option exercises and purchase accounting; net of those, share count has fallen roughly 25% over the past decade. Critically, management buys at modest multiples — typically 8-12x earnings — so the buyback math has been accretive on average. Estimated average P/IV of buybacks: ~0.6, which is good. Buffett's standard for buybacks is 'buy below intrinsic value, sized to opportunity' — Aflac roughly meets it.

5. Dividends. Aflac is a Dividend Aristocrat with 42+ consecutive years of dividend increases. Current yield ~2%. The dividend has grown roughly in line with earnings, leaving plenty of room for buybacks. Management has explicitly stated a 'dividend + buyback' return-of-capital framework with 60-70% of FCF returned to shareholders.

Communication quality. Aflac's investor day materials and quarterly disclosures are solid but not exceptional. They are transparent about Japan FX hedging, segment reporting, and capital ratios. They are less forthcoming about per-policy economics, lifetime value, and channel-mix shifts. Amos's letters lack the analytic depth of the best CEO letters (Markel, Berkshire, Watsco) but are honest about challenges (Japan demographics, US worksite competition).

The blemishes. The 2024 third-party agency fraud scandal (where contracted Japan agents falsified policies) was material, cost ~$300M in remediation, and exposed weak channel-control processes. Management responded promptly but the issue should not have grown that large. CEO Amos's compensation has occasionally drawn ISS criticism for being high relative to peer median.

Net assessment. Disciplined buybacks below IV, no value-destroying M&A, a 42-year dividend record, and prudent leverage. Lacks the truly elite reinvestment opportunities of a compounding machine because the underlying market is flat — but management has correctly returned the capital rather than chasing growth.

Capital allocator: B+

Industry Structure

Aflac competes in two distinct industries: Japanese supplemental health insurance and US voluntary worksite insurance. Porter's Five Forces analysis:

1. Rivalry among existing competitors (MODERATE-LOW in Japan, MODERATE-HIGH in US). In Japan, Aflac faces Dai-ichi Life, Nippon Life, MetLife Japan, and a handful of domestic mutuals in the cancer-and-medical supplemental segment. Aflac's ~70% share in cancer policies gives it a price-leader position; rivalry is more about product innovation (riders, lump-sum payouts) than price war. In the US, the voluntary worksite market is fragmented: Aflac, Unum/Colonial, Allstate Voluntary Benefits, MetLife, Lincoln Financial, Cigna's supplemental. Competition is steady but not destructive — premiums are small enough that price competition rarely cuts margins materially.

2. Threat of new entrants (LOW in Japan, MODERATE in US). In Japan, regulatory licensing, distribution barriers (Japan Post and bank channels are locked up), brand recognition (Aflac duck is iconic in Japan), and the unit economics of small-premium policies make new entry uneconomic. In the US, entry is easier — any state-licensed insurer can write voluntary benefits — but building the worksite agent force and broker relationships at scale takes 5+ years. Insurtech entrants (Hippo, Lemonade, etc.) have not meaningfully attacked supplemental.

3. Bargaining power of suppliers (LOW). Aflac's 'suppliers' are its agents, brokers, and reinsurance partners. Reinsurance pricing has been benign for supplemental (low catastrophe exposure). Agent commissions are bounded by market norms. Investment management is partially in-house and partially outsourced; manager fees are not material to economics.

4. Bargaining power of buyers (LOW-MODERATE). Individual policyholders have minimal bargaining power because the premium is small and they are not price-shopping. Worksite buyers (HR departments) have more leverage — they can switch carriers at renewal — but switching costs (re-enrollment, employee communication) blunt this. Group accounts above ~10,000 employees can negotiate hard; small accounts cannot.

5. Threat of substitutes (MODERATE-RISING). This is the most concerning force. Substitutes include: (a) employer-paid major medical with lower deductibles, (b) HSAs/FSAs that absorb the same out-of-pocket costs Aflac is designed to cover, (c) Medicare expansions in the US, (d) Japan's national health insurance gradually expanding cancer coverage. Each of these chips at the 'why I need supplemental' question. The rise of high-deductible health plans has actually been a tailwind in the US (more out-of-pocket = more demand for supplemental cash benefits) but the long-term policy direction matters.

Value pool location. The value pool sits in distribution-controlled, in-force premium pools. Aflac captures it because it owns the channel relationships. The pool is not growing — Japan is flat, US is growing 2-3% — but it is durable, and Aflac's share is not eroding fast.

Trajectory. Slowly shrinking in Japan due to demographics; slowly growing in the US. Net: roughly flat.

Industry Verdict: Good

The industry is neither excellent (no growth, demographic headwinds, regulatory ceiling) nor poor (high persistency, low rivalry, structural barriers). It is a 'good business in a slow industry' — which is exactly the kind of setup that produces compounders when management runs disciplined buybacks at low multiples. The verdict is 'Good' rather than 'Average' because the structural barriers (distribution exclusivity in Japan, brand in both markets) are real.

Inversion (Bear Case)

I am now a short-seller. I am paid to be right that AFL is a value trap and that owning it at $112 is a mistake. Here is the strongest case I can build.

1. The single event that kills this: the yen breaks 200 per dollar and stays there. Aflac's earnings are roughly 70% yen-denominated. The current USD/JPY around 150-155 is already historically weak; a move to 175-200 — driven by BoJ's continued accommodation, persistent US-Japan rate differentials, or a Japanese fiscal crisis — would translate Aflac's $4B+ of yen earnings into $3B or less. Book value would compress similarly. Management's hedging program covers a portion of net investment income but cannot hedge the structural earnings translation in perpetuity without burning cost-of-hedge that destroys the very economics being hedged. The market knows this. The 0.43 P/IV is not a bargain; it is a correctly-priced FX-vulnerable earnings stream. The base IV of $260 was modeled in current yen and current rates — change the rates and the IV moves to $180 or $150.

2. Why the moat is narrower than bulls think. The Japan Post distribution relationship that bulls cite as exclusive is not contractually exclusive in perpetuity — it is a commercial alliance that Japan Post can renegotiate. As Japan Post's own insurance subsidiary (Kampo) recovers from its 2019 fraud scandal and rebuilds product offerings, the incentive to keep selling Aflac through 20,000 post offices weakens. The brand 'Aflac duck' is iconic in Japan but the buyer demographic for the iconic-duck cohort is dying. Younger Japanese consumers (under 40) increasingly buy supplemental online through Hoken-no-Madoguchi and similar comparison agents — a channel where Aflac's distribution moat is zero. In the US, the worksite agent channel is being slowly disrupted by HR-tech platforms (Gusto, Rippling, Justworks) that build voluntary-benefits marketplaces directly into payroll software. Aflac is not partnered with the leaders. Five-year erosion is plausible.

3. Why management is worse than it appears. Daniel Amos has been CEO for 35 years. Long CEO tenure is a double-edged sword: he understands the business, but he also has not been pressure-tested by adversity in a generation. The 2024 Japan agency-fraud incident exposed weak channel governance — not a one-off, but a symptom of a management that has gotten comfortable. Capital allocation has been mediocre on the reinvestment front: the Zurich Japan acquisition (2017) was small, and Aflac has not redeployed its $130B investment portfolio aggressively even as Japan's interest rates began to normalize. The buyback discipline is real but the company has paid 12-15x earnings in some years (2021, 2022) — buying near intrinsic value, not at the deep discount Buffett's framework demands. Management's 'we return 60-70% of FCF' is a corporate rule of thumb, not an opportunistic capital allocator. Real compounders deploy capital when the IRR is highest; Aflac's deployment has been calendar-driven.

4. What bulls are extrapolating that won't hold. Bulls extrapolate (a) stable Japan persistency, (b) growing US worksite premium, (c) buyback share-count reduction, and (d) stable or rising USD/JPY. Each is contestable. Japan persistency is stable today but the in-force book is aging; lapse will rise as policyholders die. US worksite is growing 2-3% but new entrants and HR-tech disintermediation will compress that to flat by 2030. Buyback math depends on stock staying cheap; if the market re-rates, the buyback's value-creation engine slows. And FX is a coin flip with a fat left tail.

5. Valuation trap (multiple compression / regime change). AFL's TTM P/E of 11.72 is near its 10y average of 9.72, meaning bulls have already paid up some. If yen weakens further AND if Japan supplemental growth disappoints AND if the BoJ raises rates and Aflac's float yield rises only slowly because its bond book is locked at low coupons — three plausible coincident events — the multiple could compress to 7-8x on lower earnings, giving a fair value of $70-80. The IV range from the scorecard ($147-392) implicitly assumes mean-reversion in growth and FX. A regime change in Japanese rates or yen breaks that assumption. The reverse-DCF implied growth of -2.88% is what the market thinks; bulls argue the market is wrong. Bears argue the market is right and the IV model is over-confident in its assumptions.

If I am right, the stock could be worth $75 within 3 years.

Lollapalooza Bias Check

Biases I am importing into this analysis right now:

1. Authority bias (HIGH). The canon excerpts are dominated by Buffett's enthusiasm for insurance — GEICO's cost moat [1][6], General Re's discipline [2], the four insurance commandments [2][3]. I am pattern-matching Aflac to those narratives. But Aflac is not GEICO. GEICO is a low-cost direct auto insurer with a structural cost advantage in a price-competitive commodity market. Aflac is a distribution-controlled specialty insurer in a slow-growth, demographics-challenged geography. The Buffett insurance template fits less well than my framing suggests.

2. Anchoring bias (HIGH). The IV base case of $260.98 is sitting in the scorecard, and my entire thesis is anchored on the 0.43 P/IV ratio. If the IV model is wrong by 30% — easily possible given the maintenance-capex uncertainty noted by the scorer — the margin of safety I am claiming evaporates. I should be more skeptical of the IV anchor and more focused on simpler stress tests (book value × ROE × 10).

3. Confirmation bias (MEDIUM). I went looking for evidence that AFL is a buy and found it: low P/E, Dividend Aristocrat, Buffett-style buybacks, durable franchise. I did not aggressively search for evidence of channel disruption, BoJ regime change, or persistency erosion. The inversion section is the corrective, but it took explicit prompting from the methodology to get there.

4. Recency bias (MEDIUM). The 2024 Japan agency-fraud event is fresh, and I gave it more weight in the management section than its actual financial impact warrants. Conversely, AFL's 42-year dividend-increase streak is so old it has lost emotional salience even though it is the most evidence-rich data point for 'management is shareholder-aligned.' I am over-discounting old evidence and over-weighting new evidence.

5. Incentive bias (LOW-MEDIUM). The Compounder framework rewards completion and conviction. There is implicit pressure to issue a recommendation other than 'Hold' or 'Too Hard' because those feel like analytical cop-outs. Watching for that pressure helps; the right answer might still be Hold or Too Hard.

6. Deprival super-reaction (LOW). I do not own the stock and have no fear of missing out, so this bias is dormant.

7. Social proof (LOW-MEDIUM). Aflac is not a crowded long. If anything, the lack of social proof makes me feel slightly contrarian, which is a form of inverse social proof — feeling smart for liking an unloved name. That, too, is a bias.

Net effect. The strongest active biases are authority (Buffett-template pattern matching) and anchoring (IV ratio). I have tried to correct them by writing a strong inversion and by haircutting the IV base case in the recommendation. The conviction is medium, not high, precisely because of these biases.

10-Year Outlook

Will Aflac in 2036 be running the same fundamental business? Yes, with adjustments. Aflac will still be selling small-premium supplemental health insurance through worksite channels in the US and through bank/post/agent channels in Japan. The product will look similar — cancer policies, accident policies, hospital indemnity — though digital distribution will have grown to perhaps 30-40% of new business in both markets.

Customer base larger? Marginally. The US worksite market should grow with employment and the persistent shift toward high-deductible health plans (which increase demand for supplemental cash benefits). Japan's customer base will shrink with demographics — total in-force policy count falling 1-2%/year by the late 2020s — partially offset by product innovation (newer cancer riders, mental health, post-acute care). Net: roughly flat policy count, growing US offsetting shrinking Japan.

Profit per customer higher? Probably. Premium per policy will rise with inflation and rider attachment. Loss ratios should remain stable. Investment yield on the float will rise modestly as the bond book rolls into higher coupons (this is the most underappreciated tailwind — Aflac's investment portfolio is locked into low-yield Japanese assets that will reprice over the next decade). Profit per customer up 2-3%/year, mostly from yield, partially from price.

Moat wider? No, slightly narrower. Distribution moats erode as digital channels compete. Brand intangible erodes as younger demographics shop online. Switching costs in worksite are eroded by HR-tech aggregators. Aflac's response — own digital distribution, partner with HR-tech, push direct-to-consumer in Japan — is rational but defensive.

Single biggest threat? Yen translation risk over a long horizon, compounded by the demographic shrinkage of the Japanese supplemental market. A USD/JPY of 200+ combined with 1-2%/year policy-count decline in Japan would compress Aflac's earnings power by 30-40% over a decade. The buyback would partially offset, but the franchise economics would be smaller.

Confidence. The fundamental business is recognizably the same in 10 years. The earnings level is uncertain because of FX. The franchise is durable but not expanding. This is a 'medium confidence' compounder — not a Costco-grade certainty (Costco is 'high'), and not a tech-platform 'low' uncertainty.

CONFIDENCE: medium