Executive Compensation API: CEO Pay vs Performance

An executive compensation API should answer two questions a proxy statement makes you work for: what did this company pay its leaders, and did that pay track performance? The StockFit /api/executives/compensation endpoint returns both, parsed straight from the SEC-mandated pay-versus-performance table in DEF 14A filings, with the reported Summary Compensation Table total, the Actually Paid figure, and Total Shareholder Return side by side, each row traceable to its accession number.

To show what that data exposes, we pulled compensation for 30 large-cap companies across six sectors and charted one number that almost nobody quotes correctly: Actually Paid Compensation, the SEC's formally named "Compensation Actually Paid" (CAP) metric. The gap between what a board reports and what an executive is actually paid turns out to split cleanly by sector, and it is the cleanest read on pay-for-performance alignment in the filings. This is a data story, not investment advice.

Where executive compensation lives in SEC filings

Executive pay is disclosed in the annual proxy statement, filed as form DEF 14A, under Item 402 of Regulation S-K. The centerpiece is the Summary Compensation Table, which reports three fiscal years of pay for each Named Executive Officer (the principal executive officer, the principal financial officer, and the next three highest-paid). It breaks total pay into salary, bonus, stock awards, option awards, non-equity incentive plan compensation, the change in pension value and nonqualified deferred compensation earnings, and all other compensation.

The catch is in the stock and option awards. Those columns report the grant-date fair value of equity granted that year, computed under accounting standard FASB ASC Topic 718. It is the accounting cost of the award on the day it was granted, not the value the executive walks away with. For an equity-heavy package, the headline "total compensation" number can be almost disconnected from what the person actually earns once the stock moves. That disconnect is exactly what a second, newer disclosure was built to fix.

StockFit parses Item 402 from the XBRL-tagged proxy and returns it as JSON, so you never scrape a compensation table out of an HTML proxy. For the wider map of which SEC forms carry what, see our guide to SEC forms; for the officer roster (who holds which title, a different question than pay), see the companion executive officers API post.

Reported pay vs Actually Paid Compensation

In 2022 the SEC adopted Item 402(v), the pay-versus-performance rule, and with it a second pay number: Actually Paid Compensation. It starts from the Summary Compensation Table total, then strips out the grant-date fair value of equity and adds back the actual change in value of that equity during the year. In plain terms, it re-marks an executive's stock and options to what they were really worth over the year instead of what they cost to grant. Because it moves with the share price, it is a far cleaner read on pay-for-performance than the reported total.

NVIDIA is the textbook case. Across six fiscal years, Jensen Huang's reported total compensation stayed in a narrow band, roughly $19M to $50M. His Actually Paid Compensation, over the same years, ran from negative $4.1M to positive $344M, tracking the stock almost perfectly.

Two things jump out. First, in fiscal 2025 the reported figure was $49.9M while Actually Paid was $344.2M, because his performance stock units were re-marked as NVIDIA's shares surged (company TSR of 1,100 against a peer group at 168, indexed to $100). Second, and more surprising to most people: in fiscal 2023, when the stock fell during the year, his Actually Paid Compensation was negative. Equity that lost value is subtracted, so a down year can drive the number below zero. Reported pay can never be negative; Actually Paid routinely is.

That mechanism produces some startling single-year gaps once you scan a broad set of CEOs. A few from the data, each pulled from the company's pay-versus-performance table:

Broadcom's Hock Tan reported $2.6M in fiscal 2024 yet was actually paid $1.15B, as equity from prior mega-grants re-marked while the stock roughly tripled. Oracle's Safra Catz took a $950k salary, no bonus, and no new grant in fiscal 2025, and was actually paid $462M because her existing equity rode Oracle's run. At the other end, Chevron's and Mondelez's CEOs each posted negative Actually Paid figures in a down equity year. None of this is visible in the reported total alone, which is the entire point of the second number.

CEO pay vs performance, by sector

If Actually Paid Compensation re-marks pay to the stock, then how far it diverges from reported pay is a measure of how equity-linked a company's pay really is. Aggregate that across a sector and a clear pattern emerges. We took the average non-CEO Named Executive Officer at five companies in each of six sectors, summed reported pay and Actually Paid across each sector for fiscal 2024, and took the ratio.

Technology executives were actually paid roughly 4.9 times their reported pay; financials about 2.4 times. Utilities, healthcare, and consumer staples cluster just above 1.0, meaning Actually Paid barely departed from reported. Energy sat below 1.0: those executives were actually paid less than reported, because energy shares lagged the 2024 market and their equity was marked down. The sectors that load pay onto equity see it swing hard with the market; the cash-and-regulated sectors do not.

One honest caveat travels with this chart. A single year's ratio reflects both how equity-heavy the pay structure is and how the stock happened to perform that year, and fiscal 2024 was a strong year for tech and finance. In a down year the same equity-linked sectors would show ratios below 1.0, exactly as energy does here and as NVIDIA's negative fiscal 2023 shows at the company level. The structural takeaway holds either way: equity-linked pay moves with performance, fixed-and-cash pay does not. Fiscal years also end in different months across these companies, so each ratio uses that company's own fiscal-year stock move.

The same split, company by company

The sector averages hide a wide spread inside each group, which the per-company view makes obvious. The chart below plots all 30 companies on a log scale, colored by sector.

Broadcom's NEO group sits near 18x, an outlier driven by the same equity-remarking that put its CEO at $1.15B. NVIDIA is near 7x. The financials cluster between 1.5x and 3.3x. Energy and a few staples names (Schlumberger, ConocoPhillips, Mondelez, PepsiCo) fall below 1.0x, where Actually Paid undershot reported pay. The point is not the ranking, it is that the spread is real and traceable: each bar is one company's pay-versus-performance table, and every number can be opened back to the source filing. A normalized third-party pay table that returns a single "total compensation" figure cannot reconstruct any of this, because it never carried the Actually Paid column or the accession trail in the first place. That filing-level traceability is the same discipline behind our point-in-time data for backtesting, and compensation alignment is one of the SEC EDGAR alternative data signals a quant can extract from proxies.

Company-selected performance measures and governance flags

The pay-versus-performance disclosure carries one more under-used field: the company-selected measure, the single financial metric a board names as the most important link between pay and performance. It is a small window into what management is actually held accountable for, and it varies by business model. The /api/executives/performance-measures endpoint returns it per fiscal year:

NVIDIA ties pay to non-GAAP operating income, JPMorgan to return on tangible common equity, ExxonMobil to cash flow from operations and asset sales, and Coca-Cola to organic revenue growth. Each measure fits the way that business creates value, and reading it tells you what the board is really steering toward. Alongside it, the /api/executives/governance endpoint returns compensation-governance flags from the same proxy: whether an insider-trading policy is in place, whether equity-award grant dates are predetermined rather than discretionary, and whether award timing considers material non-public information, the disclosures the SEC added to surface spring-loading risk.

Pulling executive compensation from the API

One call returns the full pay-versus-performance history for a ticker:

curl 'https://api.stockfit.io/v1/api/executives/compensation?symbol=NVDA' \
  -H 'Authorization: Bearer $STOCKFIT_API_KEY'

Two rows from the live response, trimmed to the fields that matter. ceoTotalComp is the Summary Compensation Table total; ceoActuallyPaidComp is the Actually Paid figure; the TSR fields are indexed to $100 at the start of the measurement window:

[
  {
    "fiscalYear": 2025,
    "ceoTotalComp": 49866251,
    "ceoActuallyPaidComp": 344188027,
    "avgOfficerTotalComp": 20343288,
    "avgOfficerActuallyPaidComp": 125583913,
    "companyTsr": 1100.69,
    "peerGroupTsr": 168.46
  },
  {
    "fiscalYear": 2023,
    "ceoTotalComp": 21356924,
    "ceoActuallyPaidComp": -4118947,
    "avgOfficerTotalComp": 9941838,
    "avgOfficerActuallyPaidComp": -1364661,
    "companyTsr": 157.05,
    "peerGroupTsr": 94.13
  }
]

Every figure traces back to a DEF 14A: NVIDIA's fiscal 2025 numbers come from accession 0001045810-25-000095, the XBRL-tagged proxy filed in May 2025. Paste that into /api/filings/search-by-accession-number and you land on the exact document the values were parsed from. The free /api/filings endpoint returns the proxy filing index for any company, and /api/financials/as-reported exposes the raw as-reported XBRL for traceability, both on the free tier. The compensation, performance-measures, and governance endpoints are on the Stock and Professional plans. You can get a token from the StockFit dashboard and call it immediately, no credit card required to sign up. If you are evaluating an executive compensation API for research, the Actually Paid column and the accession trail are what separate a usable dataset from a normalized one.

FAQ

What is the difference between Actually Paid Compensation and the total?

The reported total in the Summary Compensation Table counts equity at its grant-date fair value, the accounting cost on the day it was granted. Actually Paid Compensation (the SEC's "Compensation Actually Paid") takes that total, removes the grant-date value of equity, and adds back the actual change in value of that equity during the year. So the total reflects what an award cost to grant, while Actually Paid reflects what it was really worth as the stock moved. For equity-heavy pay the two can differ by an order of magnitude.

Can Actually Paid Compensation be negative?

Yes. Because Actually Paid adds back the change in equity value during the year, a year in which the stock falls subtracts value, and the figure can drop below zero. NVIDIA's CEO had negative Actually Paid Compensation of about $4.1M in fiscal 2023, and Chevron's CEO was around negative $23.8M in fiscal 2023, both driven by equity awards losing value. The reported Summary Compensation Table total can never be negative; Actually Paid often is in a down year.

What SEC filing shows executive compensation?

The annual proxy statement, filed as form DEF 14A, under Item 402 of Regulation S-K. It contains the Summary Compensation Table (salary, bonus, stock and option awards, and more for each Named Executive Officer) and, since 2022, the pay-versus-performance table with Actually Paid Compensation and Total Shareholder Return. StockFit parses both from the XBRL-tagged proxy and returns them as JSON via the executive compensation endpoints.

What is the SEC pay-versus-performance disclosure rule?

It is Item 402(v) of Regulation S-K, adopted by the SEC in 2022 and first appearing in 2023 proxies for calendar-year companies. It requires a table relating Actually Paid Compensation for the principal executive officer and the average of other Named Executive Officers to company Total Shareholder Return, peer-group TSR, net income, and a company-selected financial measure, filed in Inline XBRL. The full SEC compliance guide is on sec.gov.

Which sectors have the biggest gap between reported pay and Compensation Actually Paid?

In our fiscal 2024 sample of 30 large caps, technology had the largest gap: executives were actually paid about 4.9 times their reported pay, followed by financials at about 2.4 times. Utilities, healthcare, and consumer staples sat just above 1.0, and energy fell below 1.0 (actually paid less than reported, as energy shares lagged that year). The pattern reflects how equity-linked each sector's pay is, so the gap widens in up markets and inverts in down ones.

Does the executive compensation API include the CEO pay ratio?

Not currently. The compensation endpoint returns the pay-versus-performance table: reported total, Actually Paid Compensation, the average for other Named Executive Officers, and company and peer-group TSR, plus the company-selected performance measure and governance flags on sibling endpoints. The CEO pay ratio (CEO pay divided by median-employee pay) is a separate DEF 14A disclosure under Item 402(u) and is not exposed by these endpoints today.

How far back does pay-versus-performance data go?

The rule took effect for fiscal years ending on or after December 16, 2022, and companies phased in three years of history initially, building toward five. In practice the StockFit data carries the pay-versus-performance table from each company's first disclosure forward, which for many issuers reaches back to fiscal 2020 or 2021 as the table accumulates years. Reported Summary Compensation Table figures predate the rule, but the Actually Paid column only exists from the pay-versus-performance era.