Chapter 6

Capital Allocation

Between FY2021 and FY2025 FactSet deployed roughly $4.1 billion of capital, and nearly six dollars in ten went to acquisitions rather than to shareholders. The defining act was the $1.932 billion purchase of CUSIP Global Services in March 2022 — a debt-funded bet that nearly doubled the balance sheet, loaded three-quarters of it into goodwill and intangibles, and roughly halved return on capital employed, from about 25% before the deal to a 13% trough and a ~20% recovery since. The franchise it bought is a genuine monopoly; the return profile it left is lower than the one it replaced.

Where the money went

Source: acquisition purchase prices from FY2022 10-K Note 6 [1] and Q2 FY2025 10-Q Note 6 [2]; buybacks and dividends from the FY2025 cash-flow statement [3].

The shape of the bars tells the story. For most of its history FactSet was a straightforward cash-return compounder: recurring subscription cash funded a rising dividend and a steady buyback. In FY2022 that pattern broke. To fund CGS, management paused repurchases — buybacks collapsed from $264.7 million in FY2021 to $18.6 million [4] — and drew down debt. Across the five years, acquisitions absorbed about $2.39 billion, buybacks about $1.0 billion, and dividends about $0.69 billion. CGS alone was 47% of everything deployed.

The dividend is the one leg that never wavered. It rose every year of the period, from $117.9 million to $160.0 million [5], about 7.9% a year, extending a multi-decade record of consecutive increases. The buyback did more work but bought less than its size suggests, a point the share count makes plainly below.

The CGS bet, decomposed

CGS is not a conventional business purchase. Of the $1.932 billion price, only $215.0 million landed in goodwill; $1.583 billion was assigned to a single intangible — the "ABA business process," a renewable license from the American Bankers Association to operate the CUSIP numbering system — with a further $164.0 million of client relationships and $46.0 million of databases [6]. Essentially the entire price bought one exclusive franchise: CGS is the sole issuer of CUSIP and CINS identifiers and the U.S. numbering agent for ISINs [7]. That is the quality case: a licensed monopoly embedded in the plumbing of the securities market, with pricing power FactSet has cited as an organic-ASV driver in every year since.

Two costs come attached. First, amortization: the acquired intangibles carry useful lives of 15 to 36 years, which works out to roughly $53 million a year of straight-line amortization on the CGS assets alone [8]. That charge depressed the FY2022 operating margin and is a standing item FactSet adds back to reach adjusted earnings [9] — part of why the GAAP-versus-adjusted gap runs the way the Financial Record describes. Second, a legal overhang: FactSet is a defendant in a purported antitrust class action over its acquisition and operation of CGS (Dinosaur Financial Group LLC et al. v. S&P Global, Inc. et al.) [10]. Management does not expect a material adverse outcome, but the suit challenges the very licensing economics that justify the price.

A precise return on the deal itself cannot be computed from the filings: FactSet does not disclose CGS revenue and calls its effect "not material" for pro-forma purposes. The one quantified marker is the auditor's control carve-out, which put CGS at 5% of consolidated revenue in FY2022 [11] — and that covered only six months of ownership. What is computable is the effect on the whole company, and it is unambiguous.

What it did to returns and the balance sheet

Source: derived from reported operating income and capital employed, FY2019–FY2025 10-Ks; operating income per the FY2025 income statement [12].

Return on capital employed averaged about 27% in the three years before CGS and fell to 13.3% in FY2022, the year the purchase price landed on the balance sheet [13]. It has climbed back to 19.8% by FY2025 as the acquired earnings season, but it has recovered only about two-thirds of the way to where it started. At ~20%, capital efficiency still sits comfortably above a ~9% cost of equity — the deal creates value at the margin — but it is a structurally lower return than the asset-light franchise FactSet was before.

The balance sheet shows why. The transaction converted FactSet from a light-capital business into an intangibles-heavy one in a single year.

Source: consolidated balance sheets, FY2022 10-K (FY2021–FY2022) [14] and FY2025 10-K [15].

Net intangibles jumped from $135.0 million in FY2021 to $1,895.9 million a year later, total assets rose about 80% to $4.01 billion, and long-term debt went from $574.5 million to $1,982.4 million [16]. By FY2025 goodwill and intangibles together were $3.20 billion — about 74% of total assets, up from 40% before CGS [17]. That concentration is less alarming than it looks, because the largest single intangible is a bona fide exclusive license rather than acquisition froth — but it does mean the reported book is now mostly the carrying value of purchased franchises, tested for impairment rather than earned in cash.

The buyback that only held the line

Source: FY2022 10-K Note 6 [18]; FY2025 balance sheet [19]; ROCE and five-year buybacks derived from reported financials, FY2021–FY2025 [20].

The five-year buyback totals about $1.0 billion, and it bought almost no share-count reduction. Diluted shares fell from 38.6 million in FY2021 to 38.4 million in FY2025 — roughly half a percent [21]. A billion dollars of repurchases essentially neutralized stock-based compensation rather than shrinking the equity; per-share compounding over the period came from the business, not from a falling count. That is not a criticism of the business — it converts cash at a high rate — but it does mean a buyer should not credit the repurchase line with per-share accretion it did not deliver.

The most recent year signals a return to the acquisitive mode. FactSet spent $243.8 million on LiquidityBook and $120.2 million on Irwin [22] [23] — order-management and investor-relations platforms bought to extend the workflow — and stepped the buyback back up to $300.5 million [24]. This reinvestment is the spending now weighing on adjusted margins, and it is the swing factor in the cash-flow growth the Margin of Safety prices.

The read

The evidence for a favorable read is concrete: ROCE at ~20% still clears the cost of capital, the acquired asset is a real exclusive-license monopoly with pricing power, leverage was rebuilt to a comfortable level within three years, and the dividend has compounded without interruption. The strongest fact against it sits in the same numbers — capital efficiency is a third lower than before CGS and has stalled around 18–20% for three years [25], the balance sheet is now three-quarters soft assets, ~$53 million a year of CGS amortization flatters the gap between reported and adjusted earnings, and an antitrust suit challenges the licensing model outright [26]. What would settle it is the trajectory of ROCE: a climb back toward the pre-deal mid-20s would confirm CGS and the recent tuck-ins are earning their cost; a stall near 18–20% while amortization and legal costs persist would mark the acquisition-led model as a lower-return version of the compounder that came before.