# Dividends vs. buybacks: the return of capital playbook
Between 2011 and 2019, IBM spent roughly $125 billion on dividends and buybacks, more than its entire market capitalization at the end of that period. The stock returned approximately 0% over the decade. Meanwhile, Visa, which IPO'd in 2008, has repurchased shares at a compound rate that has *added* an estimated 12% to per-share intrinsic value annually. Same playbook. Opposite outcomes.
This is the central tension every CFO faces when the cash pile grows beyond what the business can reinvest: how to return capital without destroying it. In 2026, with US corporates having returned over $1.4 trillion to shareholders in 2025 alone, and with the 4% excise tax on buybacks now firmly embedded in the IRA and creating real friction, the question is sharper than ever.
Modigliani-Miller told us, in a frictionless world, that the choice between dividends and buybacks is irrelevant. The real world is anything but frictionless. Three frictions dominate the modern CFO's calculus:
1. Taxation. In the US, qualified dividends are taxed at up to 23.8% (including NIIT) at the investor level, while buybacks defer taxation until the investor sells, and now incur a 4% corporate-level excise tax under the Inflation Reduction Act. For an investor base dominated by tax-deferred institutions (pension funds, 401(kkThe average number of new users each existing user generates through referrals. Above 1.0, growth compounds on itself and becomes exponential.)s), the dividend/buyback tax wedge matters less than CFOs often assume.
2. Signaling. A dividend initiation or increase is a credible commitment to sustained cash generation. Cutting a dividend is one of the most punishing actions a CFO can take, GE's 2017 dividend cut from $0.24 to $0.12, and then to a penny in 2018, vaporized $30 billion of market cap in days. Buybacks, by contrast, are discretionary. The market interprets them as management's view on intrinsic value, *if* management has credibility on capital allocation.
3. Value transfer. This is the one most CFOs underweight. Every buyback transfers value between two groups: selling shareholders and remaining shareholders. If the company buys at a price *above* intrinsic value, selling shareholders win and remaining shareholders lose. If it buys *below* intrinsic value, the opposite happens. A dividend treats all shareholders equally. A buyback does not.
This last point is what Warren Buffett hammers in nearly every Berkshire annual letter: "Buybacks make sense only if shares are repurchased below intrinsic value, conservatively calculated." Most CFOs nod and then proceed to authorize buybacks based on EPS accretion math.
Here is the trap. A CFO's banker presents a buyback model: "At our current P/E of 22x and after-tax cost of debt of 4%, a $5 billion debt-funded buyback is 6% EPS-accretive in year one." The board approves. The buyback executes.
But EPS accretion is a near-mechanical outcome whenever earnings yield (E/P) exceeds after-tax cost of debt. It tells you nothing about whether value was created. A buyback at 30x earnings is EPS-accretive if you borrow at 3%, but you've just locked in a 3.3% return on capital deployed against assets that may be worth half what you paid.
The right question isn't "is this accretive?" It's: "What is intrinsic value per share, and how does today's market price compare?"
From 2005 to 2019, IBM repurchased approximately $135 billion of its own stock. CEO SamSamServiceable Addressable Market: the slice of TAM you can realistically reach given your current business model, geography, and distribution channels.View full definition → Palmisano famously committed to a "Roadmap 2015" target of $20 EPS, and his successor Ginni Rometty inherited that promise. The math worked through aggressive buybacks: shrink the denominator faster than earnings could erode.
The problem: IBM's core businesses, hardware, services, software licenses, were structurally declining as enterprise IT migrated to cloud. AWS launched in 2006. Azure in 2010. IBM kept buying back stock at $150, $200 per share through 2013-2014, financed increasingly with debt. By 2019, long-term debt had climbed to over $60 billion, and the stock traded at $135. The company had borrowed money to retire shares at prices it would never see again.
Contrast this with what IBM *didn't* do: it made only one transformational acquisition (Red Hat, $34 billion, 2019) during the cloud era, and starved R&D relative to peers. Capital that should have gone into rebuilding the franchise went into propping up EPS.
The lesson for CFOs: buybacks are a residual, not a strategy. When buybacks become the primary EPS driver, the business is signaling that management has run out of investment ideas, and the market eventually figures this out.
Visa presents the inverse case. Since its 2008 IPO, Visa has reduced share count from approximately 2.3 billion to roughly 1.95 billion as of late 2025, a modest ~15% reduction. But here's the key: Visa repurchased shares while:
Vasant Prabhu, Visa's CFO until 2024, and his successor Chris Suh have run what is essentially a textbook program: dividends signal the floor (sustainable cash distribution), buybacks absorb excess cash that can't be deployed at the company's hurdle rate. Visa's hurdle rate is brutal, north of 20% ROIC, so the bar to *not* buy back stock is high.
Critically, Visa has been opportunistic on timing. During the March 2020 COVID crash, when the stock dipped to $135, Visa accelerated repurchases. When the stock ran to $290 in late 2024, buybacks slowed. This is what disciplined capital allocation looks like: buy more when cheap, less when expensive.
Here is the framework I recommend CFOs use when the treasurer walks in with excess cash:
Before any return-of-capital decision, ask: can this capital earn the company's cost of capital + a margin inside the business? If organic investment, M&A, or working capitalworking capitalWorking capital is the difference between a company's current assets and current liabilities, measuring short-term liquidity and the funds available to run daily operations.View full definition → deployment can earn 15%+ in a 9% WACC company, return-of-capital is the wrong answer. This is where IBM failed; this is where Visa succeeds.
The dividend should represent the cash flow you are confident the business generates *through a downturn*. A useful test: what would FCFFCFFree Cash Flow is the cash a company generates from operations after funding the capital expenditures needed to maintain and grow its asset base.View full definition → look like at trough margins from your last two cycles? Size the dividend so payout never exceeds 50-60% of that figure. This is why energy companies historically maintain modest base dividends and use variable dividends (Pioneer Natural Resources pioneered this) or buybacks for cyclical excess.
Anything above the sustainable dividend base and reinvestment needs is candidate buyback cash. But, and this is the discipline most companies lack, the buyback should be price-sensitive. Establish an intrinsic value range with the board. Buy aggressively below the low end, opportunistically in the middle, and pause near or above the high end.
Knowledge check
1. Between 2011 and 2019, IBM spent roughly $125 billion on dividends and buybacks. What was the approximate total shareholder return on the stock over that decade?
2. Under the Inflation Reduction Act, what is the corporate-level excise tax rate currently applied to share buybacks in the US as of 2026?
3. GE's 2017 dividend cut from $0.24 to $0.12 (and then to a penny in 2018) destroyed approximately how much market capitalization in the days following the announcements?
4. Select ALL correct answers about the frictions that distinguish dividends from buybacks in the real world (versus Modigliani-Miller):
Sélectionnez toutes les réponses correctes.
5. Select ALL correct answers about the signaling properties of dividends versus buybacks:
Sélectionnez toutes les réponses correctes.
Three current factors reshape the playbook:
The IRA's 1% buyback excise tax, in force since 2023 and now widely expected to rise to 4% under pending bipartisan reform proposals, materially changes the math at scale. For a company like Apple, which repurchased approximately $95 billion in FY2024, a 4% tax is $3.8 billion of friction annually. That doesn't kill buybacks, but it tilts the marginal dollar back toward dividends or M&A. CFOs should rerun their distribution mix with the higher rate as the base case.
Under the EU's Corporate Sustainability Reporting Directive, in full effect for large EU and EU-listed entities since FY2024 reporting, companies must disclose how capital allocation decisions align with stated sustainability strategy. This has real implications: returning $10 billion to shareholders while announcing a "net zero by 2040" transition that requires $25 billion in capexcapexCapital Expenditure (CapEx) is money spent to acquire, upgrade, or extend long-lived assets like equipment, property, or software that deliver value over multiple years.View full definition → invites stakeholder challenge. Several European utilities, Iberdrola, Enel, have explicitly reduced buyback activity to fund the energy transition.
The OECD's 15% global minimum tax, now substantially implemented across major jurisdictions in 2025-26, has compressed the after-tax cash generation of low-tax-jurisdiction structures. For US multinationals that previously held cash offshore at low effective rates, the post-Pillar Two cash available for return is materially lower than 2017-2022 levels. Your buyback authorization should reflect this new normal, not pre-Pillar Two run rates.
Post-2022 rate environment, where investment-grade spreads have normalized at 130-180 bps over Treasuries and BBB credits price meaningfully wider than A credits, has revived rating discipline. AT&T's $40 billion buyback program from 2014-2018, financed largely with debt, contributed to its downgrade trajectory and the eventual Warner Media unwind. Boeing's pre-737 MAX buyback binge, $43 billion from 2013-2019, left it with no balance sheet flexibility when the crisis hit. The cost of capital lesson: a buyback financed by debt is a permanent capital structure decision dressed up as a temporary capital return.
When a company believes its stock is materially undervalued and wants to deploy a large amount quickly, a Dutch auction tender (used by companies including IAC, Dell, and most recently by several mid-cap tech firms in 2024) lets shareholders bid the lowest price at which they'll sell. This concentrates capital return while letting non-tendering shareholders increase ownership. It's superior to open-market buybacks when there's a strong conviction view on undervaluation.
ASRs let a company immediately retire a large share block by paying an investment bank upfront, with final pricing determined over a subsequent averaging period. They're useful when EPS optics matter near a reporting date, but they remove the price discipline that should govern repurchases. Use sparingly and skeptically.
Pioneered in the US energy patch (Pioneer, Devon, Coterra), variable dividends have spread to other cyclicals. The structure: small fixed base dividend + formulaic variable distribution based on FCFFCFFree Cash Flow is the cash a company generates from operations after funding the capital expenditures needed to maintain and grow its asset base.View full definition →. This combines the credibility of a dividend with the discretion of a buyback, and avoids the value-transfer problem of repurchasing at peak prices. Worth serious consideration for any cyclical business.
1. Build an intrinsic value range with your board, and update it quarterly. Buybacks without a view on intrinsic value are