Skip to main content

Sum-of-the-Parts

Updated Jun 24, 2026 at 2:35pm

Research Draft High 1,176 words

Sum-of-the-parts (SOTP), also called break-up or "asset-based holding-company" valuation, prices a diversified company by valuing each business segment separately with the method best suited to it, summing those enterprise values, then adjusting for corporate-level items (net debt, unallocated costs, minority stakes) to reach equity value. Its core tension is that a company is not necessarily worth the sum of its pieces: SOTP measures theoretical break-up value, and the gap between that figure and the actual market capitalization — the "conglomerate discount" — is the whole point of the exercise, but that gap may reflect real frictions to unlocking value rather than mispricing.

How it's calculated / formed

The mechanics are an assembly of other valuation tools rather than a standalone formula:

1. Disaggregate into segments. Use the company's reported operating segments (the segment footnote in the 10-K under ASC 280 / IFRS 8 is the usual starting point) plus any separately identifiable assets — listed equity stakes, real estate, joint ventures. 2. Value each segment with the appropriate method. A high-growth software unit might be valued on EV/Revenue or EV/EBITDA from pure-play comparables; a stable industrial unit on EV/EBIT or a DCF; a real-estate arm on NAV; a listed minority stake at its market value (often haircut for tax/liquidity). Corporate Finance Institute and Wall Street Prep both stress that the multiple should come from standalone peers in that segment's industry, not the parent. 3. Sum the segment enterprise values to get a gross enterprise value. 4. Bridge to equity value. Subtract net debt, subtract the capitalized value of unallocated corporate overhead (a frequently omitted step), add non-operating assets/cash, subtract minority interest and any pension/legal liabilities, then divide by diluted shares for an implied per-share value.

A standard expression: Equity Value = Σ(Segment EV) − Net Debt − Corporate Overhead (capitalized) + Non-operating Assets − Minorities. Because each input is a range, SOTP outputs a range, not a point estimate.

How it's used in practice

SOTP is the default lens for conglomerates and holding companies — Berkshire Hathaway, Alphabet, Amazon, large industrials — where a single blended multiple is meaningless because segment risk/growth profiles diverge. It serves three distinct audiences:

  • Investors use the SOTP-vs-market gap to flag a possible "hidden value" long: if pieces are worth more than the whole, a catalyst (spin-off, divestiture, activist campaign) could close the discount.
  • Activists and bankers use it as break-up analysis — the explicit argument for splitting a company. The 2024–2025 wave of separations (General Electric's three-way split, Honeywell's announced break-up, Kellanova/WK Kellogg) were all framed in SOTP terms.
  • M&A uses it to value a target a buyer intends to dismember and sell in pieces.

The analytically honest version always pairs the gross SOTP with a bridge showing what destroys the difference: overhead, taxes on hypothetical disposals, stranded costs, and the time/risk to actually execute a separation.

Adoption, debate & evidence

SOTP is universally taught and used; what is genuinely contested is the conglomerate / diversification discount it tries to surface. Practitioner sources commonly cite discounts of roughly 10–25%, with figures like "13–15%" appearing widely — but treat these as rules of thumb, not measured constants.

The academic anchor is Berger and Ofek (1995, J. Financial Economics), who imputed segment values from median pure-play multiples over 1986–1991 and found U.S. diversified firms traded at an average value loss of roughly 13–15% (overinvestment and cross-subsidization were the cited drivers); they put the aggregate 1995-dollar loss across discounted diversified firms at about $800B. Lang and Stulz (1994, J. Political Economy) independently found diversified firms carry lower Tobin's q, and Servaes (1996) documented a large discount in the 1960s that decayed to roughly zero by the 1970s — so the discount is time-varying, not constant. The existence of a measured discount is nonetheless one of the more replicated findings in corporate finance.

The causal story is far less settled:

  • Self-selection (Campa & Kedia, 2002; Graham, Lemmon & Wolf, 2002): Campa & Kedia show that once you control for the endogeneity of the diversification decision, the discount shrinks and can even flip to a premium — firms that diversify share characteristics negatively correlated with value. Graham, Lemmon & Wolf find that diversifiers acquired already-discounted units, attributing roughly half of the measured value loss to that rather than to diversifying itself. Diversification may be a symptom, not the cause.
  • Measurement bias (Hund, Monk & Tice; book-value-of-debt critiques): because market-to-sales multiples fall as firms get larger, and diversified firms are larger, the standard imputation method can manufacture an artifactual discount tied to size rather than organizational form.

So the empirically defensible statement is: diversified firms measure as discounted on average, but how much of that is destroyed value vs. selection and measurement artifact is unresolved. SOTP tells you a gap exists; it does not prove the gap is free money.

Strengths & limitations

Strengths: SOTP is the only sensible approach when segment economics diverge sharply; it forces granular, comparable-driven thinking; it directly quantifies the spin-off thesis; and it surfaces hidden assets (a listed stake, land) that blended multiples bury.

Limitations: It is only as good as segment disclosure — many firms report deliberately coarse or shifting segments, and inter-segment transfer pricing, shared services, and unallocated corporate costs are hard to assign. The single biggest misuse is summing gross segment values and ignoring the bridge: forgetting capitalized corporate overhead, the tax leakage on hypothetical disposals, dis-synergies, and the execution risk/time-discount of actually separating the businesses. This inflates "hidden value" that may never be realizable. A second failure mode is assuming the discount will close — discounts can persist for years absent a catalyst, and a controlling shareholder or entrenched management can keep it open indefinitely.

Sources

Disputes flagged: The existence of a measured diversification discount is well replicated; its cause (value destruction vs. self-selection vs. measurement artifact) is genuinely contested. Commonly cited discount magnitudes (10–25%, "13–15%") are rules of thumb / estimates, not measured constants — do not treat as precise.