Marsh McLennan: A Wonderful Broker at an Uncomfortable Price
Opening
Marsh McLennan is the world’s largest insurance broker — a business that sits between corporate clients and insurers, designs and places risk, and collects fees and commissions for the trouble. It is the kind of company value investors tend to admire: capital-light, durable, quietly compounding. The question, as ever, is not whether the business is good. It is whether the price leaves room to be wrong. The short answer: not yet.
Gate 1 — Circle of Competence
The business model is knowable. The company earns commissions tied to insurance premiums and fees for advisory work, operating through two segments. Risk & Insurance Services — roughly 64% of revenue — contains Marsh, the world’s largest insurance broker, and Guy Carpenter, a reinsurance broker. Consulting, around 36%, contains Mercer (human resources, pension, and health advice) and Oliver Wyman (strategy consulting). In 2025 the company produced roughly $27 billion in revenue across 130 countries with 95,000 employees. Because the business requires almost no capital, free cash flow conversion is high: $5 billion of FCF in 2025, around 18% of revenue.
Information access is straightforward — a long listing history and high-quality filings — so the first layer of competence passes. The second layer, predictability, passes with a condition attached. Corporate risk — disasters, cyber, litigation, people — will exist in ten years and will be more complicated, and large-risk intermediation resembles a stable duopoly of Marsh and Aon. But part of the brokerage business rests on information asymmetry — market knowledge, insurer access — precisely the territory AI could erode. The judgment here: low-complexity brokerage (standardized small-business products) is erodible, but Marsh’s core — large-corporate and specialty risk — rests not on information asymmetry but on negotiating power from placement scale, advisory relationships embedded in client workflows, and professional accountability (brokers carry the E&O liability). Explaining this business ten years out is an exercise in analysis, not probabilistic betting — with the caveat that the judgment itself must be tested against observable indicators.
Gate 2 — Moat
The moat has three sources. First, switching costs: a large corporation’s risk program sits on years of accumulated risk data, renewal history, and insurer relationships; replacing a broker risks coverage gaps, so attrition is structurally low. Second, efficient scale and two-sided network effects: the world’s largest placement volume creates negotiating leverage with insurers and proprietary market data — global rate indices — which in turn attracts clients. A flywheel. In reinsurance, Guy Carpenter is effectively in a duopoly with Aon. Third, intangibles: a 154-year brand and a regulatory license network across 130 countries.
Evidence of pricing power is indirect but persuasive: seventeen consecutive years of adjusted operating margin expansion. Even in a soft market — global commercial insurance rates fell 5% in Q1 2026, the seventh consecutive quarterly decline — fee-based advisory growth and margin defense continued. The headwind is real, though: Marsh’s organic growth slowed to 4% in 2025 from 7% in 2024.
The moat is widening, gently. The McGriff acquisition (2024, all-cash) extended the US mid-market distribution network; emerging risks — cyber, climate, AI liability — raise the advisory intensity, and therefore the value, of brokerage; and small brokers, unable to fund AI and technology investment, are becoming acquisition fodder.
The AI question cuts both ways, and it is the central issue for this stock. On the erosion side, BofA estimated roughly $15 billion of low-complexity insurance commissions industry-wide are exposed to AI disintermediation — a concern that sent broker shares falling sharply from five-year highs in 2026. On the reinforcement side, the center of Marsh McLennan’s revenue is complex large-corporate risk plus consulting; its exclusive placement data becomes more valuable in an AI era, and its capacity to invest dwarfs smaller competitors. Management has been explicit: “Marsh is not a commodity distributor.” The net assessment: the moat’s core is reinforced by AI; its periphery — the standardized small-business brokerage the company just expanded into via McGriff and MMA — is under erosion pressure. Net effect neutral to mildly positive for now, with moderate confidence.
Gate 3 — Management
CEO John Doyle has run the company since 2023, an internal promotion with an insurance career spanning his working life. The dividend rose from $0.90 in 2012 to $3.05 in 2024 — an 11% CAGR — with another 10% increase in 2025, alongside $11 billion in cumulative buybacks.
One flag: 2024 was a record M&A year, with roughly $27 billion of deal value including the all-cash McGriff purchase — large acquisitions at a moment when broker valuations sat at historical highs. Buying at peaks is the classic capital-allocation failure. The mitigants: McGriff’s strategic fit is clear, and 2025 FCF of $5 billion, up 25%, demonstrated digestion capacity. There is no history of repeated value destruction, so this is a monitoring item, not a disqualifier.
Candor is a positive: management has acknowledged the soft-market headwind and the organic growth slowdown (7% to 4%) in numbers, and has faced the AI threat directly rather than avoiding it. Leverage is the deduction: debt-to-equity reached 1.28 funding McGriff — manageable for a capital-light broker (total debt roughly $20 billion against $5 billion FCF), but well above a Buffett-style ceiling of 0.5. Deleveraging execution should be watched.
Gate 4 — Price
Owner earnings come to roughly $4.23 billion — a conservative figure versus 2025 FCF of $5 billion, chosen to account for differences in intangibles amortization. Applying 12x–15x — treating the business as between slow growth and “certain growth,” discounted for soft-market cycles and the AI debate — yields an intrinsic value range of $50.8 billion to $63.5 billion. Generously, at 15x the 2025 FCF.
Against that: a market capitalization of $84.9 billion, a P/E of 22.3, and an owner-earnings yield of 5.0% — a 34% premium to even the top of the intrinsic value range, and only 0.7 percentage points above the 4.3% ten-year Treasury. There is no margin of safety. A 30% discount to the top of the range implies a buy-zone market cap corresponding to roughly $90 per share — a level that would realistically require either amplified AI panic or a systemic crisis. Even granting the generous 15x-FCF valuation, a 30% discount still implies about $107 per share. With the 2026 AI-driven correction in progress, tracking the price is worth the effort.
Postscript from the Watchlist
Two monitoring notes since the thesis date. On August 17, 2026, the adjusted owner-earnings yield (4.7661% at the $184.21 close) crossed just above the ten-year Treasury (4.724%) for the first time — driven by the share price falling 1.92%, not by the Treasury line moving. By August 19’s close the gap had narrowed again to −0.0126 percentage points below the line (4.6404% versus 4.6530%), mostly because the Treasury yield itself fell. Neither reading reflects any change in the business-level indicators — organic growth, margins, leverage, and FCF are unchanged. The stock remains above the valuation line; it has merely drifted closer to the Treasury line.
Verdict
Watch. Gates 1 through 3 pass: a duopoly-like position at the world’s largest broker, switching costs plus scale, seventeen straight years of margin expansion, and acceptable capital allocation with one watch item. Gate 4 fails on price alone — a 34% premium to the top of the intrinsic value range, with an owner-earnings yield offering thin cushion over Treasuries. The AI erosion question is real but currently looks like pressure on the periphery, not the core. Should panic push the market cap toward the low-$50 billions — roughly $100 per share — the case deserves a second hearing.
What would change the verdict: either a price decline into the buy zone described above, or evidence from the monitoring indicators — Marsh organic growth below 2% for two consecutive years, a break in the margin-expansion streak, AI direct-placement examples appearing in the large-corporate segment, or leverage failing to fall below 1.0 by 2027 — that turns “watch” into “reject.”
This analysis is AI-generated, educational, and not investment advice. Figures may contain errors or be delayed. Disclaimer