When the Wall Street Journal analysed roughly 2,000 proxy filings and annual reports from the S&P 1500 in the year ended June 1, 2026, it arrived at a headline number: close to $600 million in perks for roughly 15,000 executives and directors. That figure sits atop a compensation structure already marked by extreme disparity — the median S&P 500 CEO pay package reached $17.7 million in 2025, while the median employee earned $89,744, a 197-to-1 ratio. The question the data opens is not whether executives receive perks but how the system that authorises them works, and which parties in that system have the power to say no.

How the Perk Ecosystem Is Structured

The $600 million total radiates from a single hub — the boards of directors across the S&P 1500 — into five named perk categories: security, personal flights, car and driver services, health and medical services, and club memberships. A sixth catch-all accounts for much of the remaining total. Individual company spokes attach to these categories. Meta Platforms paid $22.5 million to protect CEO Mark Zuckerberg. Apple gave CEO Tim Cook nearly $790,000 in personal flights. Apollo Global Management spent $790,000 for CEO Marc Rowan’s car and driver. JPMorgan Chase spent $85,000 for Jamie Dimon’s car use. Palantir Technologies CEO Alex Karp received $160,000 in health coverage.

The map’s most important cross-link is the security justification. Apple’s board requires Cook to fly private “for security and efficiency purposes.” Meta cites “specific threats” to Zuckerberg. Apollo says Rowan’s car and driver provide “increased efficiency, convenience and confidentiality.” The same rationale appears across three otherwise independent perk categories. Equilar data show that security spending was already rising — among 208 early S&P 500 proxy filers, the prevalence of security perquisites rose 47.6 percent from 2021 to 2024 — but the 2024 killing of UnitedHealth Group executive Brian Thompson accelerated the trend.

That security justification connects downstream to shareholder scrutiny. When a board recasts personal-adjacent consumption as institutional risk management, the framing reduces the salience of the perk as a target for proxy-vote opposition. The relationship map identifies this as a dampening chain: security justification → lowered scrutiny → reduced opposition to perk approval. The mechanism is structurally implied but not directly observed in the filings; no example of a shareholder challenge blocked by a security rationale appears in the source material. Yet the absence of counter-evidence is itself consistent with the hypothesis: if the security rationale successfully dampens challenges, those challenges would not appear in public filings or shareholder votes. The mechanism is a candidate, not established — security framing plausibly shifts the debate from “does the executive need this?” to “does the company face this risk?”, a reframe that makes opposition politically costly for institutional shareholders evaluating fiduciary exposure.

The Ratio That Frames Every Line Item

Each individual perk sits atop a compensation structure already marked by extreme disparity. The $600 million in perks is not a substitute for that $17.7 million base; it is additive to it. A $790,000 personal-flight allowance at Apple, a $131,000 golf-club initiation at Texas Roadhouse, a $107,000 hunting-ranch allocation at Rush Enterprises — each reads differently when the recipient already holds a compensation package in the nine figures.

The perk data also reveals cost divergences within identical categories. At two large financial firms, car-and-driver spending differed by a factor of ten: $790,000 at Apollo versus $85,000 at JPMorgan. Both companies employ security-sensitive executives. The disparity suggests the price of a perk is set less by the underlying need than by what the board will approve — a judgment made inside a structure with limited external check.

A Benchmarking Loop With No Outside Input

That structure has a characteristic feature: pay consultants survey the perk disclosures of peer companies, produce “market-rate” benchmarks, and present those benchmarks to the compensation committees that retain them. The boards then cite the benchmarks as justification for the next round of disclosures. The loop has no external check. Aalap Shah, a pay consultant at Pearl Meyer, characterised security and health perks as “really necessary to run the business in the most effective way” and cited “an existential risk to the company if something happens to the CEO.” Rosanna Weaver, of the Interfaith Center for Corporate Responsibility, countered that “a lot of these perks are excessive” and that “there’s a whiff of entitlement here.” The two positions map onto a structural divide: one treats perks as business inputs that happen to benefit the executive; the other treats them as consumption additive to extreme compensation.

Proxy advisory firms such as ISS and Glass Lewis evaluate the aggregate compensation package, not individual perk line items. Institutional shareholders — the parties with formal voting power — tend to direct their attention at total pay levels rather than the composition of its components. The loop’s exact mechanism — how surveys are designed, how boards cite specific benchmarks in authorisation documents — is structurally asserted but not directly observed in the source filings; the pattern is inferred from the ecosystem’s architecture rather than documented in individual proxy statements.

Who Has a Voice — And Who Does Not

The parties with formal governance authority — boards and CEOs — sit as definitive stakeholders: high power, high legitimacy, high urgency. Institutional shareholders are dominant stakeholders: high power and legitimacy but low urgency on perks specifically, because the $600 million is a rounding error against the $17.7 million median CEO base. Internal heterogeneity matters here: passive index funds tend to support management on say-on-pay votes, while activist funds may push for change. The SEC, which sets disclosure rules under Regulation S-K Item 402, is a dormant stakeholder — it has the power to reshape the landscape but no current urgency.

The most structurally absent parties are retail shareholders — millions of individual investors holding S&P 1500 stock through index funds, with no organised constituency for perk-specific action — and rank-and-file employees, whose $89,744 median earnings give them the strongest equity claim on compensation fairness but no formal governance voice. The disclosure regime itself was designed for specialist consumption: the roughly 2,000 filings are publicly available but practically parsed only by proxy analysts, compensation consultants, and governance-focused investors. This information asymmetry defaults to management advantage, leaving retail shareholders with a vote on say-on-pay but no practical way to assess the individual perk line items that collectively cost $600 million.

Two absent parties carry stakes that the current governance architecture does not accommodate. Future CEO candidates — not yet in role — absorb the current perk landscape as baseline expectations; if $790,000 in personal flights and $131,000 in club dues are normalised for sitting CEOs, they become anchoring points for successor negotiations, setting future compensation floors that today’s decisions produce but no current stakeholder reviews. The general public holds a tax-expenditure stake: to the extent perks are non-taxable to recipients, the public subsidises them, but the public has no formal standing in the proxy process and exerts pressure only through indirect political channels.

Perks Without Security Cover Stand Out

The security justification does not extend to all perks. Unusual disclosures — $20,000 worth of free alcohol at Constellation Brands, a $35,000 boat allowance at Brunswick, $107,000 for personal use of a company-owned hunting ranch at Rush Enterprises, $121,502 in sports tickets at Chemed, $131,000 in golf-club initiation and dues at Texas Roadhouse — lack the security rationale. Texas Roadhouse’s board said the golf membership was intended to “better connect the CEO locally” after his relocation from Texas, a business-purpose justification that is thinner than security. These perks may be more vulnerable to scrutiny because they lack the cover that security-framed spending enjoys.

The roughly $430 million in perks beyond the security-justification bridge do not inherit it, and it is precisely their lack of institutional cover that makes them the most legible targets for the scrutiny that the security framing may deflect from the categories it covers. Other relocation items disclosed: CSX covered a $1.6 million loss on its chief operating officer’s house sale, and Kruti Patel Goyal, now Etsy’s CEO, received $1.2 million in perks covering housing, lease termination, tuition, tax preparation, and other costs during an international assignment and subsequent move to the U.S. Each is justified by its own thin business-purpose reasoning rather than the security frame.

The Security Figure That Anchors the Rest

The surge in security spending — $170 million across 250-plus companies, more than double the 2021 median — accelerated after the 2024 killing of UnitedHealth Group executive Brian Thompson. Boards reassessed their fiduciary exposure to physical risk. Meta cited “specific threats” against Zuckerberg. Apple cited Cook’s security profile. The resulting spending increase was real and, by the companies’ account, responsive to documented threats.

But security spending also functions structurally within the broader perk ecosystem. It is the largest category. It carries the most defensible rationale. And its growth establishes a precedent: when the most expensive perk category is framed as institutional necessity, the categories that follow — flights, cars, relocations, club memberships — inherit the framing by proximity, at least for those that adopt the security rationale. Security firms and private aviation providers sit outside the governance frame entirely — commercial beneficiaries of the $170 million surge, classified as discretionary stakeholders with low power and no formal voice. If security spending were to contract, these supply-side parties would lobby against reduction, adding a political dimension the current analysis does not explore. Their absence from the stakeholder map is itself a structural observation: the system has no mechanism for representing the commercial interests that depend on its continuation.

An alternative reading of the system is that perks are a second-order governance issue. At $600 million across 1,500 companies, the average per-company perk spend is roughly $400,000, or approximately 2.3 percent of the median CEO’s $17.7 million package. For passive index funds managing trillions, perk-level precision is noise. The governance question becomes not “are these perks justified?” but “why does the system generate this level of disclosure granularity for an item most shareholders treat as below their attention threshold?” The answer may be that the disclosure regime produces visibility the governance system does not process — the data exists but has no effective audience.

What the Structure Implies

The relationship map and stakeholder map together suggest that individual perk stories can be read in isolation without seeing the security-justification pattern that runs across multiple categories or the accountability mechanism it may dampen. Interventions at the board or proxy-disclosure level would affect all spokes; interventions at the security-rationale level would primarily affect the categories that have adopted that justification. Much of the remaining total beyond the five named categories remains largely unexamined, and the proxy-disclosure regime — 2,000 filings that are publicly available but practically consumed by specialists — creates an asymmetry between what is disclosed and what is visible to the general public.

A Wall Street Journal analysis of the underlying data found error rates of 2 percent or below across all categories (using string matching, dollar-weighted random sampling, and a hand review of the largest perks in each category), giving the $600 million figure a precision that most perk disclosures individually lack. The companies mentioned in the article were given the opportunity to comment.

A reader following the next wave of proxy filings might ask: Which companies will broaden the security justification to cover additional perk categories? How will institutional shareholders — particularly the passive index funds that hold most S&P 1500 shares — respond to aggregate perk growth when the next say-on-pay vote arrives? Will the SEC’s pending consideration of security-expense disclosure change the rules of the game? And in a system where the median employee earns $89,744, which perks will survive scrutiny without the shield of security?

The question the data raises is not whether executives face security threats — several clearly do. It is whether a governance structure in which boards authorise perks, consultants benchmark them against other boards’ authorisations, and shareholders evaluate the total without examining the components produces outcomes that track executive need or outcomes that track the absence of constraint. The security-justification bridge is the most analytically precise cross-link in the system — it connects disparate categories through a single defensible logic — but it covers a minority of the dollar total, and the medium-confidence link between that framing and dampened shareholder scrutiny remains a candidate mechanism rather than an established one. What a reader carries to the next story is the structural observation: the governance architecture processes detailed perk data into aggregate votes, the benchmarking loop ensures the pattern compounds, and the parties with the strongest equity claim on fairness — median employees at 197-to-1 — have no seat at the table where the decisions are made.

Analytical techniques used in this piece

This analysis applies the methods below. Each links to a short, plain-English explainer you can read and reuse.

Red-Team Assessment
Models a capable adversary probing a plan for the seams they would exploit.
Relationship Mapping
Extracts the network of ties among people, institutions, and entities.
Stakeholder Mapping
Charts the parties to a situation — their interests, power, and alignments.