$37.90
$80.94
bear $37.90
base $57.33
bull $80.94
margin of safety $33.00
market price* $119.97
Aflac sells supplemental health and life insurance (cancer, medical, accident and life policies) in Japan and the United States. Aflac Japan produced about 53% of adjusted revenues in 2025 and has more than 22 million policies in force there, about 14 million of them cancer policies. Its 2018 alliance with Japan Post Holdings is its most important sales channel in Japan. Because Aflac is an insurer, its borrowing and its policy reserves are the raw material of the business, so we value the cash it earns after interest and do not subtract its $8.3 billion of notes a second time. Interest paid was $197 million in 2025. What we measure. Owner earnings are operating cash flow, minus the swing in receivables and advance premiums (the only true working-capital line in its cash-flow statement), minus share-based pay, minus maintenance spending on property and equipment (Aflac does not report capital spending separately, so depreciation of about $36-54 million a year stands in for it). We do not strip out changes in policy liabilities, deferred acquisition costs or tax liabilities, because these are the insurance business itself. On that basis owner earnings were $7.03 per share in 2016, peaked at $8.10 in 2020, and then fell to $5.90 (2022), $5.35 (2023), $4.50 (2024) and $4.68 (2025). In dollars the fall is steeper: from $5.8 billion in 2020 to $2.5 billion in 2025, with buybacks shrinking the diluted share count from 828 million in 2016 to 535 million in 2025 and 506 million by mid-2026. Why cash flow has fallen while reported profits have not. Net earnings were $3.6 billion in 2025, well above owner earnings. The gap is mostly the change in policy liabilities, which added $2.0-2.1 billion a year to cash flow in 2019-2020 but subtracted $552 million, $302 million and $815 million in 2023, 2024 and 2025. The annual report gives two reasons. First, many Japanese policies are limited-pay products: customers pay premiums for a fixed period and remain covered for life. As they reach paid-up status, premiums stop while claims keep coming; this alone cut Japan's earned premiums by about 15 billion yen in 2025, and annualized premiums in force in yen fell 3.0% in 2024 and 2.5% in 2025. Second, claims rise as policyholders age, while maturing bonds are reinvested below the yields older policies were priced on. Earned premiums in Japan also fell about 21 billion yen in 2025 because of an internal cancer reinsurance arrangement with Aflac Re, which moves premiums between group companies and does not change the group's economics. Reported earnings are also lifted by large reserve remeasurement gains (about 68 billion yen in 2025), which are accounting estimates rather than cash. We therefore trust the cash figure over the earnings figure. The starting level. We use the median of 2023-2025, $4.68 per share (about $2.5 billion), because those are the years in which the paid-up drag has been fully in place; the earlier, higher years describe a block of business that was still mostly paying premiums. Growth. In 2025 owner earnings per share rose 4.0%, but only because of buybacks; total owner earnings fell 1.6%. Our base case starts with a 2% decline in year 1, below both figures. First-half 2026 operating cash flow was $1,080 million against $988 million a year earlier (+9%), and the policy-liability line turned positive (+$431 million against -$195 million). Two quarters are encouraging but not enough to move the starting level. We give no credit for new product launches, sales growth, the Japan Post alliance producing more, a yen recovery or further buybacks: none is contracted or a proven, repeatable source of cash growth in recent years. The aging and paid-up profile of the Japanese block is, by contrast, a structural feature of the business visible in the filings, so it is carried in every case. Credit risk. The investment portfolio carries real credit exposure: about $3.6 billion of floating-rate transitional real estate loans, $4.3 billion of middle market loans (mostly below investment grade), $1.4 billion of commercial mortgages, $4.1 billion of limited partnerships and a book of below-investment-grade bonds, against shareholders' equity of $30.3 billion at June 2026. In 2008-2009 Aflac took heavy losses on perpetual bank hybrids and European financial securities and had to shore up Aflac Japan's solvency ratio. That record is the model for our bear case. The three cases, valued at a flat 6%, after interest.
Base case, $57.33 per share: owner earnings start at $4.68 and decline 2% a year for ten years, to about $3.82, as the paid-up and aging drag continues at roughly its recent pace, then keep fading 2% a year. Business worth $57.33 per share; no debt is subtracted, so equity value is $57.33.
Bear case, $37.90 per share: owner earnings decline 5% a year, the pace of the paid-up and yen-premium erosion if it persists at 2024-2025 rates without any 2026 relief, plus one credit cycle like 2008-2010. We charge about $1.5 billion of after-tax credit losses spread over years 2-4 (about $1 per share a year), sized from losses of roughly 10% on the transitional real estate loans, 8% on middle market loans and a repeat of the hybrid-security write-offs of 2008-2009. Owner earnings bottom near $2.81 in year 4 and fade 5% a year thereafter. Even at the trough, cash flow covers interest more than seven times, and shareholders' equity of about $60 per share would absorb the losses without the need to raise capital, so we do not model dilution. Equity value $37.90, which is below book value.
Bull case, $80.94 per share: the policy-liability drag, which averaged about $556 million a year in 2023-2025 (about $1.04 per share), ends as the first-half 2026 figures suggest, lifting owner earnings to about $5.70 per share, held for ten years, then fading 2% a year as the block continues to age. Equity value $80.94. What changed since our previous analysis (late September 2026: $33.10 bear, $57.01 base, $76.26 bull). The starting cash level is unchanged. The base case is essentially the same: we now assume a gentle continuing decline rather than a flat decade followed by a fade, and the two shapes produce almost the same value. The bear case is a little higher because it is now built from a specific credit-cycle loss on the loan book plus a steady 5% erosion, rather than an open-ended 8% annual decline that exceeded what the paid-up mechanism explains. The bull case is a little higher because the improvement in the policy-liability line now has two quarters of support and fades more slowly after year ten. What we tested. The bear case rests on the policy-liability drag continuing. Against it: first-half 2026 operating cash flow rose 9% and the policy-liability line turned positive. For it: annualized premiums in force in yen fell for a second straight year, and the annual report says claims payments are expected to rise over the life of a policy. Both facts are reflected in the spread between the cases. Margin of safety price. We are moderately confident in the earnings level: three years of cash data agree within about $0.40 per share, but the cash trend has been falling for five years and depends on actuarial reserving we cannot fully see into. The worst plausible outcome is a credit shock in the real estate and middle-market loan books arriving together with a sharp fall in the yen, which would cut dividends from Aflac Japan to the parent. The structure would survive it: debt is modest at about $8.3 billion against $30 billion of equity and cash flow covers interest more than ten times, there is no single customer, and the business does not rely on a controlling owner. If we are wrong about the paid-up drag easing and about credit losses, the shares are worth something close to the high-$30s. During the holding period, the business currently pays out about $2.3 per share a year in dividends, which supports the return. From $33 a share the bear case implies about 15% upside plus a dividend yield of roughly 7%, and the base case implies roughly a doubling over a few years, so $33 is the price at which a 20% annual return is implied even if the bear case unfolds within two to three years. Above that, the implied return depends on the policy-liability drag stopping, which is not yet proven. What would make this wrong: renewed double-digit declines in operating cash flow, a further steep fall in the yen, or credit losses on the transitional real estate loans beyond the 2008-2010 experience would move value toward or below the bear case; a full year of positive policy-liability cash flow would move it toward the bull case.