A headline of "CEO pay: $70M" makes most people picture one big deposit hitting a bank account. That's not what happens. The SEC's 2024 data put the median S&P 500 CEO total compensation at $16.3M — but on average only about 11% of that is cash. The other 89% is stock and options that vest conditionally over the next several years. Three transformations sit between the headline and what eventually lands in the executive's account: accounting valuation, time, and tax. Let's walk through each one using the actual filings — SEC DEF 14A in the U.S., Korea's DART system — that generate the numbers.

1. What the headline really is — Grant-Date Fair Value

Every U.S. public company files a Proxy Statement (DEF 14A) that contains a seven-column "Summary Compensation Table." Almost every CEO pay headline you've ever read traces back to that table. Here are its columns.

1. Salary. Annual cash base. S&P 500 median ~$1.3M (Equilar 2024).
2. Bonus. Discretionary cash bonus — mostly $0 at large caps, with (5) taking its place.
3. Stock Awards. RSUs and PSUs valued at Grant-Date Fair Value. S&P 500 median ~$9.4M.
4. Option Awards. Stock options valued via Black-Scholes — really "future potential value," contingent on the stock clearing the strike price.
5. Non-Equity Incentive Plan. Cash bonuses for hitting short-term targets like revenue or EPS. Median ~$3.2M.
6. Pension and NQDC. Change in pension value and nonqualified deferred compensation.
7. All Other Compensation. Aircraft, security, car, the rest. Median ~$0.4M.

Here's where everyone gets fooled. Columns (3) and (4) are valuations at grant, not amounts actually received. If the stock drops before RSUs vest, realized pay comes in far below the headline; if it climbs, far above. And an option is worth exactly nothing if the stock stays below strike.

2. Grant-Date vs realized — three sharp examples

The SEC's August 2022 "Pay Versus Performance" rule now requires DEF 14A filings to disclose both (a) the standard Grant-Date Fair Value and (b) the "Compensation Actually Paid" — the value of awards that actually vested or paid that year. Three cases show just how far the two can drift apart.

Case 1 — Tim Cook (Apple), FY 2022. Summary Compensation Table headline: ~$99.4M. Realized value of PSUs that vested the same year: ~$99.1M. Practically on top of each other, because Apple's stock held steady. Honestly, this close a match is the exception, not the rule.

Case 2 — Mary Barra (GM), FY 2022. Headline of $29M. Then GM's stock fell 41% that year, and the realized value of the awards that actually vested came in around $15M — about 52% of the headline. The decline flowed straight through to realized pay.

Case 3 — Elon Musk (Tesla), 2018 options package. This is the wild one. In January 2018, Tesla's board granted Musk 12 tranches of performance-tied options, each unlocking on a specific market-cap, revenue, or EBITDA milestone. The Black-Scholes value at grant was about $2.6B. By 2022, with all the milestones met, the realizable value had ballooned to roughly $52B. Delaware's Chancery Court invalidated the package in January 2024, and Tesla shareholders re-ratified it in June 2024. One person's pay package ended up generating corporate-law precedent.

3. CEO-to-worker pay ratio — Dodd-Frank §953(b)

The 2010 Dodd-Frank Wall Street Reform Act §953(b) requires every U.S. public company to publish, annually, the ratio between the CEO's total compensation and the median employee's. Since the 2017 first filing year, that "Pay Ratio" has become one of the most cited CEO-pay metrics around.

AFL-CIO Executive PayWatch's 2024 data put the average S&P 500 ratio at 268:1 — meaning CEOs at large caps earn, on average, 268× the median worker. The Economic Policy Institute's 1965 estimate was 20:1, so the multiple has grown about 13× over six decades.

A few cases worth a look.

• Costco — FY2024 ratio ~245:1. Relatively low, because the median worker (part-timers included) earns around $50K.
• Tesla — the 2018 ratio hit a record ~40,668:1 thanks to the one-time recognition of Musk's options package. Normal years are unremarkable.
• Amazon — 2024 ratio ~6,474:1, pulled up by low fulfillment-center median wages.

The ratio's limits are obvious. Global firms with low-wage overseas labor look worse on paper; consulting and tech firms concentrated at headquarters look better. It's hard to compare companies on a single number, so proxy advisors ISS and Glass Lewis treat it as a secondary signal, not a headline.

4. Say-on-Pay — the shareholder veto

The same Dodd-Frank Act, in §951, introduced Say-on-Pay: an annual (or triennial) advisory shareholder vote on executive compensation. The vote is non-binding. But a defeat sends the compensation committee back to the drawing board in practice.

ISS's 2024 report put average S&P 1500 Say-on-Pay approval at ~89% — mostly passing — while about 4% of companies failed (under 50%). Two notable 2024 failures.

• Intel — failed in May 2024 with 35% approval, on shareholder pushback over a $179M one-time RSU grant to incoming CEO Pat Gelsinger.
• Boeing — failed in April 2024 with 36%, after $32.8M went to departing CEO Dave Calhoun while the 737 MAX crisis was still unfolding.

The mechanism has produced steady, incremental pressure. Companies with repeat failures tend to drift toward higher PSU weights, tighter option vesting, and shorter CEO tenures.

5. Korea — what DART does differently

Korea's 2013 amendment to the Financial Investment Services and Capital Markets Act required listed firms to disclose individual compensation for registered (등기) executives earning more than ₩500M. The disclosure lives in each annual report (사업보고서) on the Financial Supervisory Service's DART system.

From 2024 annual reports:

• Lee Jae-yong (Samsung Electronics) — stepped off the registered board in 2024, so his individual compensation no longer falls under the disclosure mandate.
• Chung Euisun (Hyundai Motor Group) — combined Hyundai/Kia/Mobis ~₩12.2B (FY2024 reports, summed).
• Park Jeong-won (Doosan Group) — disclosed ~₩8.7B in 2024.
• Koo Kwang-mo (LG Group) — the LG Corporation report shows ~₩7.1B.

Four structural differences from the U.S.

1. Unregistered executives sit outside disclosure. The U.S. mandates disclosure for "named executive officers" defined by function; Korea draws the line at board registration, which can route some chaebol heads right out of the public dataset.
2. Stock and option weights are far lower. Korean registered-exec pay is dominated by salary plus bonus, with RSUs and options typically under 20% — versus ~70% at U.S. large caps.
3. No pay-ratio disclosure. Korea has no §953(b) equivalent.
4. No Say-on-Pay. Korean shareholders vote on aggregate director-pay caps at AGMs, but never on individual packages.

6. Bebchuk and Fried — the Pay Without Performance debate

Harvard Law's Lucian Bebchuk and Jesse Fried codified the "managerial power theory" of executive pay in their 2004 book Pay Without Performance. The core claim is simple: boards nominally set CEO pay, but because CEOs influence who sits on the board, the negotiation runs mostly one direction.

Three structural arguments they cite.

1. Pay for luck. When an entire industry booms (oil prices rising, say), oil-major CEO pay grows mechanically — but doesn't shrink proportionally in the bust.
2. Camouflage. Spreading compensation across options, pensions, and retirement packages lowers shareholder visibility compared with plain cash.
3. Stealth compensation. Personal loans, jet usage, family travel, and similar perks tucked inside "All Other Compensation."

The book started a real academic fight. Kevin Murphy (2013) and others fired back with a "market competition theory" — that globalizing CEO labor markets naturally drove top-talent prices up. Both theories are partly right. Neither one is fully sufficient on its own.

7. A five-step checklist for reading a headline

1. Start with the source. Read the actual SEC DEF 14A or DART filing, not a Forbes summary. U.S.: sec.gov/edgar; Korea: dart.fss.or.kr.
2. Break apart the seven-column table. Separate cash, stock, and options.
3. Put Grant-Date next to Compensation Actually Paid. The Pay Versus Performance section shows both — a wide gap signals heavy stock dependency.
4. Check the Pay Ratio. §953(b) disclosure; line it up against industry medians.
5. Track Say-on-Pay approval. Three-year trend. Anything under 80% signals board-shareholder tension.

8. Where the CEO Salary Quiz trips players up

Two failure modes show up over and over.

Mistake 1: "Famous company, so the pay must be huge." Costco's Craig Jelinek earned about $9M in 2024 — among the lowest at his revenue tier. Costco has a cultural norm that CEO pay should reflect the same frugality the company sells. Size correlates with pay, but it isn't linear.

Mistake 2: "This must be the normal year." A new CEO's sign-on package — stock to replace awards forfeited at a prior employer — can balloon a first-year headline. The number drops back to a normal level the following year. Skip the check for a one-time grant, and you'll read the trend completely wrong.

The same trap shows up in the media. "CEO X pay up 80% year-on-year" is often just a sign-on grant in the base year. One sanity check on whether a one-time award explains the jump goes a long way.