demographic model · gilded age fortunes · 2026
A bottom-up projection of how many people today are economically or institutionally connected to fortunes that cemented during the Gilded Age (1870 - 1914). The model starts from 76 root families, simulates descendants over six generations with documented fertility and extinction patterns, overlays marriage alliances, and classifies the living class into core controllers, peripheral inheritors, and symbolic heirs. All numbers are scenario ranges, not a census.
headline finding
With the long-tail of regional dynasties, maternal-line descendants who lost the surname, and the in-law network folded in, the model places the living class between 3,449 and 30,018 people across the conservative-expansive bracket (central 10,113). Of that, only roughly 39 to 98 occupy core institutional controller roles.
On wealth, the visible-only reading is misleading and the model now states the full range explicitly. Documented identifiable holdings are $300B - $600B in 2026 dollars — but the cloaking apparatus catalogued in the article (dynasty trusts, single- and multi-family offices, foundations, LP positions, LLC chains, donor-advised funds) is designed precisely to keep wealth out of those registries. A bottom-up cloaking estimate puts total class holdings at $1.3T - $3.4T (realistic), with an aggressive-cloaking band reaching $5.3T - $12.6T, and a specialized-rate compounding ceiling at $37.9T. Weighted by the source-anchored probability distribution below the cohort chart (which places mass on the cloaking-honest bands because the founding families invented the apparatus rather than adopting it), the central reading is $14.7T — roughly 53% of 2026 US GDP.
At the visible endpoint, the implied 1913 - 2026 realized real return is only 4.07% — below the long-run 6.7% standard-market real rate. At the cloaking-honest realistic mid, it is 5.60%; under aggressive-cloaking assumptions, 6.86%; at the specialized ceiling, 8.23%. Weighted by a source-anchored probability distribution that accounts for the founding families' role as architects (not adopters) of holding companies, private foundations, family offices, and dynasty trusts, the expected wealth is $14.7T — implying a 1913 - 2026 realized real rate of roughly 7.33% , modestly above market. The apparent underperformance in the visible record is an artifact of measuring only the visible portion.
On the West Coast specifically: the genuine Gilded-Age Western cluster is not centered on Beverly Hills (a post-1920 enclave) but on the SF Bay Big-Four-railroad and Comstock-silver lineages (Crocker, Stanford / Huntington remnants, Hearst SF roots, Spreckels, Mackay, Flood, Fair), plus the LA-area Doheny / Chandler-Otis families and the San Marino / Pasadena corridor that Henry E. Huntington effectively built. The seasonal-compound geography (Pebble Beach 1919, Sun Valley 1936, Montecito since the 1900s) was itself a Gilded-Age development. With the Comstock fortunes added to the root list and Pebble Beach / Montecito added to the transport map, the West Coast represents roughly 11-14% of the modeled class, anchored by the Pacific-Union Club, Burlingame Country Club, California Club, and Bohemian Club.
methodology
Enumerate Gilded Age families whose peak fortune exceeded roughly 0.05% of US GDP at the time, or who held controlling stakes in strategic industries (rail, steel, oil, banking, real estate). Record founder cohort, surviving children, branches by ~1920, and institutional legacy.
For each family, run a Monte Carlo over six generations (1870 - 2050 cohorts) using empirically grounded fertility means and per-generation extinction draws. Family-specific extinction risk biases the per-step Bernoulli.
Estimate intra-elite marriage rates per generation, accumulate cross-family edges, then deduplicate descendants whose two parents both trace to the root list. The resulting graph yields a cluster count and a hub-share metric.
Split living descendants into core controllers (board chairs, principal trustees, family-office principals), peripheral inheritors (substantial trust beneficiaries / foundation board members), and symbolic heirs.
Add married-in spouses and the extended in-law and named-beneficiary network that aggregates into trusts, foundations, family offices, and holding LLCs. Report a separate count of beneficial owners obscured behind those vehicles.
scenario ranges
Parameters that drive the spread: average completed children per couple per generation, per-generation lineage extinction probability, intra-elite marriage rate, per-tier core controller seats, peripheral share of remaining descendants, spouse multiplier, and cloaking factor.
collective wealth
Sum of peakNetWorthShareOfGdp across the explicit root
list equals 20.7%
of contemporaneous (1913) US GDP. Anchoring on US nominal 1913 GDP
of $40B and applying a 40% founder-cohort
philanthropy haircut yields ~$5B
of effective starting capital that entered the compounding bath in
1913. The counterfactual compares passive standard market returns
against a specialized PE / family-office vintage premium over 113 years, with a 55% lifetime leakage applied (estate
tax, divorce, dispersion, scandal, bad investment, consumption).
Using documented holdings of $0.3T - $0.6T implies a 1913-2026 realized rate of only 4.07% — below the standard market rate of 6.7%. But the article's own cloaking taxonomy says dynasty trusts, family offices, foundations, LP positions, and LLC chains are designed precisely to keep wealth out of those registries. Treating the visible portion as the truth is internally inconsistent with the rest of the model.
Adding a defensible bottom-up cloaking estimate of $1.0T - $2.8T brings the total to $1.3T - $3.4T (realistic combined). The implied 1913 - 2026 realized rate at the probability-weighted expected value of $14.7T is roughly 7.33% — modestly above the 6.7% standard market rate. Under aggressive cloaking the rate rises to 6.86%; at the specialized ceiling, 8.23%. The "structural underperformance" reading was a measurement artifact of looking only at visible registries.
At the realistic combined endpoint, the class controls roughly 8% of 2026 US GDP. Under aggressive cloaking, ~ 32%. The probability-weighted expected value is $14.7T ≈ 53% of US GDP — comfortably above the entire Forbes 400 (~19%), roughly comparable to the top 0.001% visible tier (~29%), and approaching the top 0.01% (~57%). The class did not lose its position; it engineered the appearance of having lost it.
cohort wealth comparison
The Gilded-Age class is plotted as a probability distribution across the five wealth bands. Each bucket carries explicit source anchors (table below the chart). The mass shifts toward the aggressive-cloaking and mid-to-ceiling bands because the founding families did not adopt private-equity / family-office vehicles — they invented their historical antecedents: Standard Oil Trust (1882, first holding company), Russell Sage Foundation (1907, first US private foundation), Bessemer Trust (1907, first formal family office), the Du Pont corporate trust, and Andrew Mellon's regime design as US Treasury Secretary 1921-1932. Multi-generational proto-PE expertise plus 113 years of compounding plus political influence on the tax / regulatory environment makes the visible-only and bottom-up endpoints structurally implausible as centers. Probabilities sum to 100%. Comparison cohorts are the Fed DFA top-end household tiers and the Forbes 400 at their documented aggregate wealth.
Anchored on visible registries. Carries low probability because the article's own cloaking taxonomy describes the very vehicles designed to keep wealth out of these registries. Taking documented holdings as the truth contradicts the rest of the model.
Documented + a defensible bottom-up estimate of cloaked holdings (single-family offices, dynasty trusts, LP positions, LLC chains, donor-advised funds). Treated as a probable floor rather than a center, because bottom-up reconstruction inherently misses what the apparatus is designed to hide.
If hidden vehicles aggregate to 3-5× the bottom-up estimate (i.e., the apparatus is more total than reconstructable). Matches independent SCF / Forbes / Zucman cross-comparison estimates that top-end wealth surveys undercount by 25-50%.
Implied if dynastic-trust opacity is near-total AND the class captured significant excess returns through proto-PE vehicles that they themselves originated. Bessemer Trust (Phipps, 1907) has approximated ~9% real returns over 100+ years on disclosed segments — a documented case of sustained specialized-rate compounding at the family-office level.
The 113-year compounded ceiling at 9% real with 55% leakage. Carries non-trivial weight because the founding families ARE the historical reference cases for sustained family-office compounding (Mellon, du Pont, Rockefeller arms, Phipps via Bessemer, Pritzker offshoots). Capped at 16% because aggregate-class realization (vs. individual-family) is constrained by extinction, dispersion, philanthropy, and political tax shocks (estate-tax peaks 1916, 1941, 1976).
For a 10,113-person cohort, the expected aggregate wealth is $14.7T (53% of 2026 US GDP) — above Forbes 400 ($5.4T, 400 individuals), above the top 0.001% ($8T, 1,300 households), and approaching the top 0.01% ($16T, 13,000 households). The class meaningfully overlaps with all three benchmark cohorts at the individual level, so these are not disjoint totals; the class is a structurally invisible component of the visible top-end.
Expected per-capita wealth is roughly $1M per class member. That sits between the top 0.001% (~$6.2B per household at 1,300) and the top 0.01% (~$1.2B per household at 13K) — i.e. the class as a whole has density comparable to the top 0.001%-to-0.01% range. The Forbes 400 remains the densest visible cohort (~$13.5B per individual), but Forbes 400 individuals are also disproportionately class members.
58% of the probability mass sits in the aggressive-cloaking and mid-to-ceiling bands ($8.5T - $20T centers). 16% sits at the specialized ceiling. Only 26% of mass sits below the realistic-combined-mid band. The distribution reflects two compounding facts: the cloaking apparatus is documented to exist, and the founding families invented the apparatus rather than adopting it.
generational pyramid
class composition
root families
Tier 1: peak fortune ≥ 0.6% of contemporaneous US GDP. Tier 2: 0.2% -
0.6%. Tier 3: < 0.2%. Top 12 by central-scenario living-descendant
count shown below; full 76-family list is the
source of truth in src/data/elite-class/root-families.ts.
machinery · NYC · founded ~1870s
Isaac Singer fathered 24 children with multiple partners; massive descendant network including European aristocratic branches.
shipping · NYC · founded ~1860s
Fortune diluted across many branches in 2-3 gens; bloodline survives, name capital largely cultural.
food · Chicago · founded ~1890s
Eleven children produced unusually large descendant pool for the tier.
rail · Twin Cities · founded ~1890s
Twin Cities anchor; large branch count; substantial Minnesota institutional footprint.
machinery · Atlanta · founded ~1890s
NC textiles, sister fortune to Cone; multi-generational regional anchor.
banking · NYC · founded ~1880s
Lazard brothers + David-Weill family; transatlantic banking; large descendant network.
banking · NYC · founded ~1880s
Bank failed 2008; family separate from bank since 1980s; descendants extensive.
banking · NYC · founded ~1880s
Goldman-Sachs intermarriage; bank long since partnership-then-public; descendants extensive.
real-estate · NYC · founded ~1840s
Splits into US and UK (peerage) branches by 1900; both lines extant.
machinery · Atlanta · founded ~1890s
North Carolina textiles; sisters Etta and Claribel donated major Matisse/Picasso collection.
banking · NYC · founded ~1880s
Eight brothers; international banking; declined relative to peers after WWI.
chemicals · Wilmington · founded ~1820s
Largest documented surviving descendant pool of any US Gilded-Age family (estimated 3-4k cousins).
sector composition
alliance network
Approximate connected-component count after accumulating intra-elite marriage edges across all generations. Ranges from 10 (expansive — more consolidation) to 17 (conservative — fewer cross-family ties survive).
Cross-family marriage edges accumulated across six generations. Each edge represents a couple in which both spouses descend from different root families on the list.
Tier-1 fortunes attract a disproportionate share of cross-family marriages. Rockefeller, Vanderbilt, Astor, du Pont and Carnegie sit at the densest end of the network even though the Carnegie blood line is thin.
Multiplier applied to the raw cross-family sum to avoid double-counting individuals who descend from two root families simultaneously. Lower factor = more historical consolidation.
asset cloaking mechanisms
Each row catalogs a legal or financial vehicle that separates legal title from effective control. The "beneficiaries per vehicle" column is the empirical range across better-documented case studies (IRS Form 990-PF for foundations, Schedule 13D filings for public-equity positions, deed records for real-estate chains, family-office directories for managed-account counts).
Holds family assets across generations with limited public disclosure; trustee is the legal owner.
GST-exempt trusts sitused in SD, NV, DE, or AK with perpetual or 360-year horizons.
Manages investments, real estate, philanthropy, and personnel for one extended family.
Bessemer Trust (originally Phipps), Rockefeller & Co (now public), Pitcairn Trust.
Bundles services for multiple lineages, aggregating allocation and obscuring individual stakes.
BBR Partners, Brown Brothers Harriman, Lombard International.
Holds and grants capital under 501(c)(3); family controls board and grantmaking.
Mellon, Rockefeller, Duke, Kellogg, Heinz Endowments.
Donor controls grant flow without operating a foundation; sponsor holds legal title.
Fidelity Charitable, Schwab Charitable, Vanguard Charitable accounts.
Family capital allocated through GP entities; LP positions hide individual contributors.
Pritzker family's Marmon, Walton Enterprises, family-anchored funds-of-funds.
LLCs nested in LLCs hold land and buildings; beneficial owner rarely in public record.
NYC, Aspen, Palm Beach, Hamptons titleholders organized as DE / NV / WY LLCs.
Splits income and remainder interest between family and charity; reduces estate tax.
CRUTs, CLATs commonly used to transfer concentrated stock at low gift-tax cost.
geographic dispersion
Upper East Side · Greenwich CT · Bedford NY · Hamptons · Tuxedo Park · Oyster Bay
Highest density of Gilded Age root families and surviving family offices.
Beacon Hill · Brookline · Manchester-by-the-Sea · Naushon Island · Pride's Crossing
Brahmin lineages: Cabot, Lowell, Forbes, Saltonstall, Higginson.
Bryn Mawr · Gladwyne · Villanova · Chestnut Hill
Old Quaker and Pew-adjacent wealth; quieter institutional footprint.
Sewickley · Fox Chapel · Wilmington Greenville · Chateau Country
Carnegie / Mellon / Frick / Phipps / Heinz / du Pont anchor.
Lake Forest · Winnetka · Kenilworth · Glencoe · Gold Coast
McCormick, Field, Armour, Swift, Pillsbury legacies extend into MN.
Pacific Heights · Atherton · Hillsborough · Woodside · Pebble Beach
Big Four railroad descendants plus modern-era overlap.
San Marino · Hancock Park · Bel Air · Pacific Palisades · Montecito
Huntington / Doheny lineage anchors; broad later overlay from migrating wealth.
Palm Beach · Manalapan · Jupiter Island · Hobe Sound
Winter enclaves built by Flagler-era development; high overlap with NY metro families.
Aspen CO · Jackson Hole WY · Sun Valley ID · Big Sky MT
Secondary residences; high concentration of family-office LP capital.
Winston-Salem · Pinehurst · Camden SC
Duke / Reynolds / Cone / Cannon families; some Hampton/Long Island migration.
London (Mayfair / Belgravia) · Paris · Geneva · Monaco · Sydney (smaller)
Astor UK peerage branch + Warburg + portion of Pritzker / Mellon expat lines.
College towns, small-city civic anchors, second-generation diaspora
Long tail of descendants no longer in primary enclaves.
probable zones · transport economy
Mapping the class onto its physical footprint: 16 private-aviation airfields, 7 urban primary clusters, 6 estate-zone corridors, 14 seasonal compounds, 3 large-landholding ranches, and 11 international outposts, joined by 14 dominant migration corridors. Aggregate annual class-tied private-aviation movements across the top airfields ≈ 255,950; the listed corridors account for ~70,800 class trips/year.
Class-tied movements ≈ (FBO annual GA movements) × (estimated class share). Teterboro tops the list because it is the default departure point for Manhattan-based class members regardless of destination.
Estimated class-tied trips per year on each major route. NYC ↔ Hamptons in summer and NYC ↔ Palm Beach in winter are the densest single corridors in the dataset.
aggregate
Implied per-class-member transport spend (central): ~$ 1078K / year averaged across the 10,113-person expanded class. The actual distribution is sharply skewed toward the ~640 core controllers and peripheral inheritors who do the bulk of private-aviation and yacht movement; symbolic heirs and synthetic long-tail members are below the median.
institutional footprint
Harvard · Yale · Princeton · Columbia · MIT · Stanford · Chicago · Penn
Trustee rolls and named-gift dockets show clear concentration of root-family surnames across multiple generations of board service.
Brown Brothers Harriman · Bessemer Trust · Pitcairn Trust · Glenmede · Pacific Family Office
Several survive as direct lineal institutions; board / partnership rolls remain partially family-staffed.
MoMA · Whitney · Frick Collection · Morgan Library · Field Museum · Carnegie Hall · BSO
Founding-family seats on governance boards remain common 100+ years after the original endowment gift.
Open Space Institute · The Nature Conservancy · National Audubon Society · Land Trust Alliance
Old real-estate families frequently appear; private-estate land donations often have governance carve-outs.
Rockefeller · Mellon · Duke Endowment · W.K. Kellogg · Pew · Hewlett · Packard
Several largest by AUM trace directly to Gilded Age fortunes; board composition partially family-tied.
Council on Foreign Relations · Brookings · AEI · Aspen Institute · Conference Board
Board memberships overlap with major family offices and asset managers connected to root families.
Cravath · Sullivan & Cromwell · Davis Polk · Simpson Thacher · Cleary Gottlieb
Trust, estate, and family-office work concentrates root-family legal needs in a small number of firms.
Groton · St. Paul's · Choate · Andover · Exeter · Hotchkiss · Lawrenceville
Multi-generational family enrollment patterns; trustee rolls overlap with university and bank boards.
cohort positioning
Mapping the 10,113-person expanded class onto identifiable institutional seats. The footprint is concentrated: roughly 1,098 governance positions (university trustees + boarding-school trustees + corporate / foundation / cultural board seats) and 12,325 class-tied memberships across 30 founder-tier private clubs, with substantial overlap (a single class member typically holds 2-4 club memberships). Annual cohort flow into the institutional system is ~704 new positions and admissions per year.
30 clubs across 11 metros. Class-share = estimated fraction of total members traceable to the modeled class. Aggregate class-tied seats = 12,325.
Aggregate ~57 class-tied trustee seats and ~205 traceable legacy admissions per year across 12 institutions. Total trustee board capacity = 446 seats; class share ≈ 13%.
~149 traceable class taps per year across the old-line Yale senior societies, Harvard final clubs, and Princeton Bicker eating clubs. These are the highest per-seat concentration of class membership in the entire dataset (Bones / Porcellian / Ivy all ≥ 35% class share).
Founded by William Huntington Russell; canonical old-line society.
Sister "old line" society; financial / legal lineage slant.
Third "Big Three" Yale society.
Aggregate of mid-tier senior societies; lower class density than the Big Three.
Oldest final club; founder-tier admissions, multi-generational legacy.
Old-line; FDR was a member.
JFK era; founder-tier identity.
JFK was tapped; Lowell legacy seat density.
Aggregate of newer / lower-density final clubs.
Oldest and most selective Bicker club.
F. Scott Fitzgerald era; founder-tier identity.
Bicker club; lower founder-density than Ivy/Cottage.
Bicker club; broader admissions.
10 schools enrolling ~1,403 students from class families at any time (out of 6,990 total enrollment), plus ~56 class-tied trustee seats. Graduating class flow per year ≈ 351.
Aggregate ~985 class-tied positions across six channels. The dynasty-controlled-firm and private-bank channels remain the densest per-seat concentrations; the Fortune 500 directorship pool is large in absolute terms but dilute (~4-5% of seats).
Surviving direct Gilded-Age governance: Hershey Trust (Hershey Co), Dorrance trust (Campbell), Mars-adjacent Wrigley, Cargill, du Pont via Chemours/Corteva, Reynolds family at Z. Smith Reynolds Foundation, Anheuser-Busch family holdings post-InBev.
Brown Brothers Harriman partners, Bessemer Trust (Phipps), Pitcairn Trust, Glenmede (Pew), Whittier Trust, Northern Trust private wealth.
Roughly 4-5% of Fortune 500 director seats; concentration in finance, media, REITs, consumer staples, and family-anchored firms.
Family offices anchor LP capital for Blackstone, Apollo, KKR, Carlyle; GP-side board interlocks more limited but present at smaller alternative managers.
Rockefeller, Mellon, Duke Endowment, Pew, Kellogg, Carnegie Corp, Hewlett, Packard, Heinz Endowments, Z. Smith Reynolds, Hershey Trust.
Met / MoMA / Whitney / Frick / Morgan / Field / Carnegie Hall / BSO / Boston MFA / Huntington Library / National Gallery / Walters.
At the top: ~12-15 "founder-tier" senior society / final club / elite city-club seats per year per institution, with class share of 35-50%. At the base: ~200 legacy admits per year across all peer universities, plus ~350 boarding school graduates per year from class families. The institutional reproduction system converts ~700 class admissions per year into the next generation of governance seat-holders.
~1,098 governance seats divided by ~10,113 expanded class members equals ~11% — i.e. about one in nine class members occupies a formal institutional position of power. The rest hold trust beneficiary interests, family-office service relationships, and social network access rather than direct seats.
Founder-tier clubs (Knickerbocker, Philadelphia, Somerset, Brook, Bath & Tennis, Bailey's Beach) show 45-60% class share — the highest concentration of class membership in the entire dataset. University boards have shifted toward meritocratic / geographic / political balance since the 1970s; club admissions have not.
limitations
what to read into the numbers
Model source: src/data/elite-class/model.ts and src/data/elite-class/root-families.ts. The simulation is
deterministic via a seeded PRNG so the headline numbers in this
article are reproducible at build time. Adjusting parameters in the
scenario blocks regenerates every figure on this page.
