Public Asset Manager Executive Compensation: methodology

Snapshot proxy-2026-09-04-9adaefde · 35 companies · 280 people · 1305 executive-year records · fiscal 2018–2026 · data compiled September 4, 2026

What this is

Named executive officer compensation at publicly traded asset managers, read from each company's proxy statement (SEC Form DEF 14A) and standardized by Michael Oak Advisors. The file behind this page holds 1305 executive-year records for 280 people at 35 companies (15 traditional asset managers, 11 alternative asset managers and 9 diversified financial companies with an asset-management arm), fiscal years 2018 to 2026. Every figure on the page is computed in the browser from those records; nothing is typed into the page.

Which records enter a benchmark

A record enters a benchmark when three things hold: the executive held the role for the full year and the compensation is completely disclosed (Present-Full and Complete in the review); the executive runs or leads the asset-management business rather than a wider financial company; and the role was mapped from the disclosed title with high confidence. 816 of 1305 records qualify; 489 do not: 241 non-investment executive at a diversified company, 135 partial year or incomplete disclosure, 94 board chair rather than an operating role, 19 role mapping unconfirmed (medium or low confidence). Records that are not benchmarked stay visible on the People tab and in each company's profile with their reason, and never enter a percentile.

Role mapping. Each disclosed title is classified against a written taxonomy of 14 roles (ruleset proxy-roles-2) by a language model given the company's whole officer list, checked against a hand-labelled sample of 77 titles it must reproduce, with Michael Oak Advisors' own decisions (42 titles) applied over it. Of 401 distinct titles, 354 were classified at high confidence, 9 at medium and 0 at low. Records whose title was classified at medium or low confidence, 25, are flagged "role mapping unconfirmed" and held out of benchmarks until they are reviewed: Head of Distribution 9, Head of Business Line 5, Other NEO 4, CFO 4, President 2, COO 1. As a result 1 role (President) has no benchmark row in the default view; their records are on the People tab. The workbook's own role column carries the same classification; 0 records differ in role and 0 in relevance. For 28 titles that carry no function (Executive Vice President, Senior Managing Director and the like) the classifier was also given a public description of the holder from the company's own site or a bio service, and the source of each is kept with the map.

CIO / Head of Investments. This benchmark row holds two taxonomy roles: the firm-wide chief investment officer (the title that says global, group, head of investments, or plain chief investment officer where no more senior investment title exists) and the head of one investment group (the chief investment officer of one asset class, the head of equities, credit, private equity, multi-asset or a portfolio-management group, or the chief executive of an investment subsidiary). Most public asset managers name no chief investment officer among their proxy officers, so the two are benchmarked together; the People tab and each company's profile show which a record is. The chief executive of a diversified company's asset-management business is a CEO, not an investment head.

Rows folded and rows never benchmarked. Chief risk, human-resources and administrative officers are too few among named executive officers for rows of their own and report in Other NEO; a chief administrative officer takes the COO row only in a year the company names no chief operating officer, and co-chief operating officers stay together as co-chief executives do. Executive and vice chairs are shown in People and company profiles but never benchmarked: a board chair is not an operating role, and a chair who is also chief executive is a CEO already. Director pay is not in the executive table. 94 records are held out for that reason.

Presidents. A president or co-president named beside a chief executive is the firm's second in command, not a comparable job across firms, so the record reports in Other NEO unless the public record sorts the person into a better row: the head of an investment platform to CIO / Head of Investments, a president who runs the business day to day with no chief operating officer to COO. President keeps a row of its own only where a company-year names no chief executive.

Companies left out of the peer set. CTBI: A Kentucky bank holding company filed in the master as a Traditional asset manager: its named officers are the bank's chief executive, chief financial officer and regional bank presidents. Mike's call 2026-09-04 afternoon after the evidence pass. The master workbook keeps the rows; the page, the download and the audit totals leave them out. 1 company is excluded.

Diversified companies. At a bank, broker or advisory firm the named executive officers mostly run the wider company; those 273 records are shown but never benchmarked. Only the executives the workbook identifies as running the asset-management business enter, and the peer-set control says so beside the business model.

The default benchmark takes each company's latest fiscal year in the file and the benchmark-eligible executives in it: 121 records at 32 companies. Years are not pooled: a company with six years of disclosure is one peer, not six. A single fiscal year is the alternative, and every year remains on the People tab.

3 companies contribute no record to the default cohort because none of their latest-year executives is benchmark-eligible: Evercore Inc.; The Goldman Sachs Group, Inc.; Houlihan Lokey, Inc. (all 3 are listed on the Companies tab and in People with each record's reason).

One record per person and year. Successive proxy statements restate the same fiscal year, and the file the page is built from held 71 such restatements, usually under the title the executive held at the later filing. The proxy filed for that fiscal year carries the title held in the year, so that row is kept. WisdomTree, which filed under two tickers, is one company.

What the dollar figures are

The Summary Compensation Table's own columns, each a pay basis, never merged. SEC Total is the default: it is the figure every company must report and the one the SEC defines. It is neither realized pay nor a consulting measure of total direct compensation. Bonus and non-equity incentive are distinct columns: most asset managers report their annual incentive under non-equity incentive plan compensation, and a company that reports it as a bonus will show the opposite pattern.

Pay mix uses only records whose seven Summary Compensation Table columns sum to the reported Total within 1%: 1290 of 1305 records, 121 of the 121 in the default cohort. On the rest the extraction's components and its Total do not reconcile, so no share is computed from them. Shares are shown as a five-part grouping (salary; annual cash incentive, which is bonus plus non-equity incentive; equity, which is stock plus option awards; pension and deferred compensation; other) and column by column as filed. The medians are taken separately, so they need not total 100%.

Company size and revenue

AUM is the figure the company reported for that fiscal year, in billions, joined from its annual report; 1058 of 1305 records carry one. A company's size for the filter is its latest fiscal year with a figure, labelled with that year, so the size of a company whose recent years lack a figure is an older one. No figure in any year: Houlihan Lokey, Inc. (shown as unavailable; it passes the size filter only when no bound is set).

One company-year carried AUM in the wrong unit in the source (The Carlyle Group Inc. FY2021, 5 records): a value above one million is read as full dollars and one between twenty thousand and one million as millions, and the band is recomputed from the corrected figure. Revenue is the source's figure divided by one million: its column is labelled millions but holds full dollars, verified against two companies' XBRL revenue on EDGAR to the dollar; 1089 records carry it, in the CSV and company detail only.

Floors, counts and the percentile

A median needs at least 2 records; a quartile, a 90th percentile or a distribution needs 5. Floors count records, not companies; the number of people and the number of unique companies are shown on every row, and a row resting on fewer than 5 companies is marked with a dagger. Percentiles are linear interpolation on position (n−1)·q, the definition Excel's PERCENTILE.INC uses; the workbook's own Dashboard reproduces under it.

Where a company discloses two chief executives in one year, both are records in the CEO row and the company counts once. In the default cohort: Blue Owl Capital Inc.; Hamilton Lane Incorporated; KKR & Co. Inc. (co-chief executives).

The pay-rank ladder

The Benchmarks tab groups the same records two ways. By role is the default. By pay rank regroups them: Rank 1 is the highest paid among benchmarked executives at one company in one fiscal year, rank 2 the next, and so on, on whichever pay basis is in use. It is the same population as the role ladder -- 121 records at 32 companies in the default cohort -- regrouped, which is what makes the two views comparable. Neither is a seniority order: rank is what a company paid, not what it called the job.

A company's top-paid officer can be absent from every rank. Rank is taken over the benchmarked records only, so a record that is partial-year or incompletely disclosed, that runs the wider company rather than its asset-management business, whose role mapping is unconfirmed, or that is a board chair takes no rank at all -- 489 of 1305 records in the file are outside every benchmark for those reasons. In the default cohort the highest rank 1 record is Stephen A. Schwarzman at Blackstone Inc. (fiscal 2025, $125,641,824). That is the highest paid among benchmarked, not the highest paid disclosed.

The ladder is published to rank 5. Ranks past it are computed and withheld: 3 records at 3 companies, a sample too thin to publish and one whose median no longer falls below the rank above it. A rung that out-earns the rung above it is worse than no rung. Across ranks 1 to 5 the default cohort's medians fall at every step. Under a narrower peer set they need not: companies disclose different numbers of executives, so a lower rung can rest on a handful of records at large payers and sit above the rung over it. The page marks a row resting on fewer than 5 companies, and the floors apply to a rank row exactly as they do to a role row.

Plan design

The Plan design tab counts practices across the peer set from each company's latest proxy statement in the data: whether an independent compensation consultant is used (6 no, 29 yes), the long-term vehicles named, whether a clawback policy and ownership guidelines are described, and the say-on-pay approval where the proxy states it (24 of 35 companies). Compensation consultants are counted as used or not used, never by name. Each company's profile carries the plan text as extracted from its proxy by Michael Oak Advisors with AI assistance; that text has not been independently re-verified and the filing governs.

Next refresh

The next refresh is expected in August 2027: the first day of the month after the last company's expected next proxy statement, one year on from its latest filing on EDGAR. EDGAR was asked on September 5, 2026: 0 of 35 companies have filed a proxy statement newer than the one their records come from, so the data is one proxy season behind today. 30 of the 35 companies file in the main season, February to May, the last of them on 2026-04-29; the rest file in December, January or July.

Sources and what is not here

Every record names the filing it was read from and its date, and every company links to its proxy statements on EDGAR; no document-level link is on file. Nothing on this page is aged to a common date, converted to realized pay, adjusted for one-time or special awards, or estimated by a model. Director compensation, the CEO pay ratio, pay-versus-performance disclosure, grant-level detail and equity ownership are in the workbook and are not published in this release.