# Optimal capital structure: debt, equity, and the real world
In April 2013, Apple's then-CFO Peter Oppenheimer did something that bewildered every finance professor in America. With $145 billion in cash sitting on the balance sheet, more than the GDP of Hungary, he issued $17 billion in bonds. It was the largest non-bank corporate bond deal in history at the time. Over the next decade, Apple would issue more than $130 billion in debt while sitting on a cash pile that routinely exceeded $200 billion. The company had no operational need for the money. It could have funded any project, any acquisition, any buyback program with a phone call to its treasury team.
So why borrow?
The answer lies at the intersection of tax arbitrage, repatriation mechanics, signaling theory, and a clear-eyed view of what Modigliani and Miller got right, and what they deliberately left out. For any CFO making capital allocation decisions in 2026, understanding the Apple playbook is non-negotiable. The interest rate regime has reset, OECD Pillar Two has rewritten the tax calculus, and equity markets are punishing balance sheet sloppiness in ways they didn't five years ago.
This lesson reconstructs the optimal capital structure framework from first principles, then shows you how the world's most sophisticated treasury teams are actually using it today.
In 1958, Franco Modigliani and Merton Miller published a paper that earned them both Nobel Prizes and frustrated generations of practicing CFOs. Their proposition: in a world without taxes, bankruptcy costs, agency costs, or information asymmetry, the value of a firm is independent of how it's financed. Debt or equity, it doesn't matter. The pie is the same size regardless of how you slice it.
Of course, Modigliani and Miller knew this was nonsense in practice. The point of M&M wasn't to describe reality, it was to identify the *frictions* that make capital structure matter. Every one of those frictions is a lever a CFO can pull.
1. The Tax Shield. Interest is tax-deductible; dividends are not. At a 21% U.S. federal corporate rate (plus state taxes pushing the effective rate to roughly 25-26%), every dollar of interest expense generates roughly 25 cents of tax savings. For a company with $10 billion in debt at 5%, that's $125 million annually in pure tax shield value. Capitalized at the cost of debt, that's a multi-billion-dollar gift from the IRS.
2. Bankruptcy and Distress Costs. The flip side. As leverage rises, so does the probability of financial distress, and distress is expensive even before bankruptcy. Customers defect (would you sign a five-year contract with a company that might not exist?), suppliers tighten terms, key employees update their LinkedIn profiles. Bed Bath & Beyond's 2022-2023 collapse showed that distress costs can vaporize 40-60% of enterprise value before the bankruptcy filing.
3. Agency Costs. Debt disciplines management. Michael Jensen's 1986 free cash flowfree cash flowFree 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 → hypothesis argued that mandatory interest payments prevent empire-building. This is precisely why private equity uses leverage so aggressively, it forces operational discipline.
4. Information Asymmetry and Signaling. When management issues equity, the market assumes management thinks the stock is overvalued. When management issues debt, it signals confidence in future cash flows. This is why equity issuances typically cause a 2-3% stock price drop on announcement, while debt issuances are largely neutral.
The modern synthesis, the trade-off theory, says the optimal capital structure balances the tax shield against expected distress costs. Graphically, it's an inverted-U: value rises with leverage until distress probability starts to bite, then falls.
In 2026, the optimum has shifted. With investment-grade yields hovering in the 5.0-5.5% range (versus 2.5-3.0% in 2020), the after-tax cost of debt is meaningfully higher than during the ZIRP era. A company that was optimally levered at 3.5x EBITDAEBITDAEBITDA (Earnings Before Interest, Taxes, Depreciation, and Amortization) measures a company's operating profitability before financing and accounting decisions, used to compare core performance across firms.View full definition → in 2021 might be optimally levered at 2.5x today. This isn't theoretical, Verizon, AT&T, and other capital-intensive issuers have spent 2024-2025 actively deleveraging, with Verizon CFO Tony Skiadas explicitly targeting a 2.25x net leverage ratio by year-end 2026.
Return to Apple. By 2017, the company had issued nearly $100 billion in bonds while holding $268 billion in cash and marketable securities. CFO Luca Maestri (who took over from Oppenheimer in 2014) was effectively running a hedge fund inside a consumer electronics company. The bond issuances had a clear rationale that combined four overlapping motivations.
Motivation 1: Cash was trapped overseas. Pre-2017 tax reform, repatriating foreign earnings triggered a 35% U.S. tax bill. Roughly 94% of Apple's cash was held by Irish subsidiaries. To fund domestic shareholder returns, Tim Cook had committed to returning capital aggressively after Carl Icahn's 2013 pressure campaign, Apple could either pay the repatriation tax or borrow domestically. At 3% pre-tax (roughly 2% after-tax), debt was vastly cheaper than the 35% repatriation hit.
Motivation 2: The tax shield was real money. Apple's cumulative interest expense from 2013-2023 exceeded $25 billion. At a 21% marginal rate, that's $5+ billion in tax savings, pure value creation.
Motivation 3: Locking in cheap long-dated money. Apple's 2020 issuance included 40-year bonds at 2.55%. With inflation now running 2.5-3% and Fed funds at 4.25-4.50% in 2026, those bonds are economic gifts to Apple's shareholders.
Motivation 4: Optimizing the cost of capital. Adding debt to a 100% equity-financed balance sheet reduces WACC even when the company doesn't "need" the money. Apple's WACC sits around 9-10%; without leverage, it would be 100-200 basis points higher. On a $3 trillion market cap, that's billions in incremental enterprise value.
The Trump tax reform's deemed repatriation and territorial system eliminated motivation #1. But Apple kept issuing debt. Why? Because motivations 2, 3, and 4 still applied, and because by 2018-2021, interest rates had fallen further. Apple issued $14 billion in bonds in 2021 at a weighted average yield under 2%.
The 2026 lesson: Apple's capital structure decisions were never about needing capital. They were about *optimizing the cost of capital* and *exploiting tax and rate arbitrages* that would have been malpractice to ignore.
Knowledge check
1. In April 2013, despite holding approximately $145 billion in cash, Apple issued $17 billion in bonds. What was notable about this transaction?
2. According to the lesson, what was the actual purpose of the Modigliani-Miller (1958) capital structure irrelevance proposition?
3. OECD Pillar Two, referenced in the lesson as reshaping CFO capital structure decisions in 2026, primarily establishes what?
4. Select ALL correct answers about the frictions that M&M deliberately abstracted away from in their 1958 proposition:
Sélectionnez toutes les réponses correctes.
5. Select ALL correct answers explaining why Apple continued issuing debt while holding massive cash reserves:
Sélectionnez toutes les réponses correctes.
Theory and case studies are useful. But on Monday morning, you need a process. Here is the framework used by sophisticated treasury teams at companies like Microsoft, Johnson & Johnson, and Unilever.
Pick a target rating before you pick a leverage ratio. The cost-of-debt curve is non-linear, moving from A to BBB might cost 50 bps, but BBB to BB costs 200+ bps because you cross the institutional investment-grade boundary. Microsoft maintains AAA largely for strategic optionality (the ability to issue $50B+ in 72 hours for an acquisition like Activision), not because it minimizes WACC. For most large-cap industrials, a strong BBB+ or A- is the sweet spot.
The more cyclical your EBITDAEBITDAEBITDA (Earnings Before Interest, Taxes, Depreciation, and Amortization) measures a company's operating profitability before financing and accounting decisions, used to compare core performance across firms.View full definition →, the lower your sustainable leverage. Compare two companies with identical $1 billion EBITDAEBITDAEBITDA (Earnings Before Interest, Taxes, Depreciation, and Amortization) measures a company's operating profitability before financing and accounting decisions, used to compare core performance across firms.View full definition →: a utility (Duke Energy, NextEra) can carry 5-6x leverage; a semiconductor company (Micron, Applied Materials) shouldn't exceed 1.5-2x at the trough. Micron's CFO Mark Murphy explicitly targets a "fortress balance sheet" through cycles, the company carries net cash going into downturns precisely because its EBITDAEBITDAEBITDA (Earnings Before Interest, Taxes, Depreciation, and Amortization) measures a company's operating profitability before financing and accounting decisions, used to compare core performance across firms.View full definition → can swing 80% peak-to-trough.
If you have abundant high-return projects (think NVIDIA in 2024-2025), preserve financial flexibility. If you're a mature cash generator (think Altria, IBM), distribute aggressively. The classic Donaldson pecking order, retained earnings first, debt second, equity last, still holds for most situations.
In 2026, refinancing risk is back. Companies that termed out 10-year debt at 2% in 2020-2021 are now facing 5%+ refinancing walls in 2030-2031. Sophisticated CFOs are running "refinancing wall" analyses and pre-funding selectively. Boeing's $10B bond issuance in October 2024 at yields of 6-7% was painful but necessary, the alternative was a rating downgrade and a much worse 2026 maturity wall.
This is the new wrinkle. OECD Pillar Two's 15% global minimum tax, now fully implemented across the EU, UK, Japan, Korea, and most major jurisdictions in 2026, has compressed the value of aggressive tax structuring. Interest deductibility games that worked in 2018, Luxembourg holdcos, Irish IP boxes, hybrid instruments, now trigger top-up taxes. The tax shield is still real, but it's been narrowed. Your treasury team needs to model the tax shield under Pillar Two, not under the legacy regime.
Capital structure decisions don't end with debt-to-equity ratios. Once you've optimized the right side of the balance sheet, you need to return capital efficiently.
Buybacks are flexible, tax-efficient (for shareholders, who pay only on realized gains), and signal management's view that the stock is undervalued. The Inflation Reduction Act imposed a 1% excise tax on buybacks in 2023, and the Biden-Harris-era proposals to raise it to 4% died in Congress, but the political risk remains. Companies like ExxonMobil, which executed $20B in buybacks in 2024, factor this into the buyback-vs-dividend mix.
Dividends signal sustainable cash flow but create a commitment problem. Cutting a dividend is a market-moving event (see Intel's 66% dividend cut in February 2023, which preceded a 12% stock drop). Once you initiate, you're locked in.
The modern playbook, used by Meta and Alphabet's debut dividends in 2024, is a *small* dividend (signaling discipline) combined with *large, variable* buybacks (preserving flexibility). Meta announced its first dividend at $0.50/quarter in February 2024 while simultaneously authorizing $50 billion in bu