SOTP, Breakup, and Asset-Based Methods

Comps and DCF share a quiet assumption: the company being valued is one business, with one risk profile, one natural peer set, and one earnings stream that a single multiple or discount rate can price. The methods in this article exist for the companies that break that assumption. Sum-of-the-parts, breakup analysis, and stub value analysis handle companies that are really several businesses reported under one ticker. Net asset value handles companies whose assets are the value, rather than a platform for earnings above them. Dividend discount models handle companies, chiefly banks and regulated utilities, where dividends are the only equity cash flow that can be projected with confidence and the standard unlevered framework does not apply at all.

None of these methods introduces new machinery. SOTP runs comparable company analysis several times, once per segment. NAV runs a discounted valuation asset by asset. The DDM is a DCF whose cash flow is the dividend and whose discount rate is the cost of equity. What changes is the unit of analysis, and recognizing when to change it is the skill these methods test.

How Sum-of-the-Parts Valuation Works

Sum-of-the-parts (SOTP) valuation, known interchangeably as breakup or break-up valuation, prices every segment of a diversified company on its own terms, with the methodology and peer group that fit that segment, and adds the pieces together. A conglomerate spanning aerospace, chemicals, and software cannot be priced on one EV/EBITDA multiple: each division carries its own growth profile, margin structure, and valuation paradigm, and a blended consolidated multiple undervalues the strongest segments while flattering the weakest.

The finished SOTP is compared against the company's current market value. When the SOTP value sits above the market value, the gap is the conglomerate discount; the opposite case, a conglomerate premium, is rare.

The Four Steps

1. 1.Identify the segments. The company's segment reporting in the 10-K or annual report supplies revenue and EBITDA (or operating income) by division. A segment only earns its own line in the analysis if it is genuinely distinct economically: a separate industry, a separate customer base, a separate growth trajectory.
2. 2.Select a segment-appropriate methodology. An industrial division is valued on EV/EBITDA against industrial peers. A software division takes EV/Revenue or a premium EV/EBITDA multiple against tech peers. A financial subsidiary takes P/TBV or P/E against FIG peers. A pharma pipeline takes risk-adjusted NPV.
3. 3.Value each segment independently. Every segment gets its own pure-play peer group of comparable companies in its industry; the comps mechanics are unchanged, only run at segment level.
4. 4.Sum and adjust. Add the segment enterprise values, subtract unallocated corporate costs, subtract net debt, and add non-operating assets such as equity-method stakes.

The final step compresses into one expression:

Unallocated corporate overhead, the G&A that belongs to no segment, is typically capitalized at a corporate-level multiple and deducted as a negative value block. Equity-method investments that sit outside the segments come in at fair value.

The Segment Data Problem

Public disclosure rarely gives you everything the analysis wants. ASC 280 requires US companies to report revenue and operating income by segment, but not segment-level balance sheets, CapEx, or working capital. In practice that means the disclosure supports a comps-based SOTP (peer multiple applied to segment EBITDA) but usually not a full segment-level DCF. The standard workaround borrows ratios from each segment's pure-play peers: if aerospace comps spend 4-5% of revenue on CapEx, assume the segment does too, and estimate D&A the same way.

Three further complications recur in almost every SOTP:

• •Shared cost allocation. Conglomerates allocate shared services (IT, legal, finance, HR) to segments on arbitrary bases such as revenue or headcount. A segment reporting a 20% operating margin under the parent's allocation might run at 15% standalone (it loses cheap shared services) or at 25% (the allocation overcharges it). Standalone margins have to be estimated, not read off the filing.
• •Transfer pricing. When segments sell to each other, internal prices may not reflect market terms; inter-segment revenue must be restated to market-based pricing.
• •The pure-play problem. SOTP is only as good as the segment comps. When a segment is genuinely unique, no true pure-play peer exists, and the analyst must use a broader comp set and apply a judgmental premium or discount to the peer median.

A Worked SOTP

Take a diversified industrial with three segments, each valued on the multiple its own peer group trades at:

From the segment total, four corporate-level adjustments produce equity value:

• •Unallocated corporate G&A of $300 million per year, capitalized at 10x: minus $3.0B
• •Net debt: minus $8.0B
• •Equity-method stake in a joint venture, at fair value: plus $1.5B
• •Estimated one-time separation costs: minus $0.5B

Netting those against the segment total gives the equity value:

On 300 million diluted shares, the SOTP implies roughly $101 per share. If the stock trades at $78 (a $23.4 billion market cap), the implied discount is ($101 - $78) / $101, about 23%, and an activist can argue that separation would unlock roughly $23 per share, or $7 billion in total.

One note on the $0.5B deduction: that line captures one-time separation costs. Recurring dis-synergies, the ongoing standalone cost increases discussed below, would instead reduce segment EBITDA or be capitalized at a multiple, exactly like the corporate overhead line; a one-time cost is deducted once at face value.

The example also shows why SOTP finds hidden value. The software segment generates only 13% of consolidated revenue but contributes $10 billion, roughly a quarter of the SOTP value, because standalone it would command 20x. Inside the conglomerate, that value is diluted by the 8x chemicals business and buried under whatever blended multiple the market applies to the whole.

Mixing Enterprise and Equity Multiples

When one segment is valued on P/E or P/TBV, the multiple produces equity value directly, not enterprise value, because price-based multiples are already net of that segment's financing. Net debt is therefore subtracted only from the EV-valued segments; the P/E-valued segment's equity value is added afterwards. Deducting corporate net debt from a total that includes a P/E-derived value double-counts the financial segment's liabilities, a classic SOTP error.

The Conglomerate Discount

The conglomerate discount measures how far a diversified company's market capitalization sits below its SOTP value, expressed as a percentage of the SOTP. Research from BCG, Credit Suisse, and academic studies puts the typical discount at 10-15%, stretching to 20-30% for companies with especially diverse or poorly performing portfolios.

At a 15% discount, the market is paying 85 cents on the dollar of separated value. The discount is not a communication failure that better investor relations could close; it reflects structural disadvantages of housing unrelated businesses under one roof:

• •Investor specialization mismatch. Institutional money organizes by sector. An aerospace fund wants aerospace exposure, not chemicals, so the conglomerate's all-or-nothing structure shrinks the natural buyer base for the stock. Sector index funds and ETFs sharpen the problem: a company straddling three GICS sectors fits none of them cleanly.
• •Capital allocation opacity. The CEO and board run an internal capital market that investors cannot see or influence. A division able to reinvest at 20% returns can be starved while cash flows to a politically favored legacy business.
• •Management dilution. No leadership team is expert in three unrelated industries at once; strategy, operating model, and regulatory knowledge all differ. Standalone companies get undivided focus and faster decisions.
• •Valuation difficulty. Because no single framework fits, many investors fall back on a blended multiple that mismeasures every segment, and shared cost allocations obscure each division's true standalone economics.

Not every discount signals value destruction. A conglomerate run by a genuinely skilled capital allocator (Berkshire Hathaway is the standing example) can add value by moving capital to its best use faster than public markets would. The analytical question is rarely whether a discount exists; it nearly always does. The question is whether it marks genuine inefficiency or nothing more than the market's taste for focused companies.

From Discount to Breakup

Breakup analysis is the SOTP put to work: does separating the company create value once the costs of separation are counted? A measured discount is the starting point, not the conclusion. Feasibility turns on four offsets:

• •Dis-synergies. The mirror image of merger cost synergies: separation recreates duplicated costs. Each standalone entity needs its own IT infrastructure, CFO, general counsel, HR, audit, and compliance; smaller entities lose purchasing scale and may borrow at wider spreads. Dis-synergies typically run 2-5% of the separated segment's EBITDA and are deducted from the SOTP value.
• •Stranded costs. Corporate overhead that does not disappear the day a segment leaves.
• •Credit impact. If the parent's rating subsidizes borrowing costs for every division, smaller and less diversified standalones may be rated lower and pay more for debt.
• •Tax friction. The structure of the separation determines how much of the theoretical value unlock survives, which is worth its own treatment.

Separation Structures and Tax

A tax-free spin-off under Section 355 distributes subsidiary shares to existing holders as a dividend, creating two independent public companies with no tax at the parent or shareholder level. Qualification requires that the business has been actively operated for at least five years, that both entities remain active businesses after separation, and that the transaction has a genuine business purpose beyond tax avoidance. When available, this route delivers the full SOTP value with no leakage.

A taxable sale of a segment triggers capital gains tax on the sale price over tax basis. For a segment worth $5 billion with a $1 billion basis, the $4 billion gain generates roughly $840 million of tax at a 21% federal rate. That friction must be netted against the SOTP value, and it partly explains why some conglomerates rationally live with their discount: a 10% discount against estimated breakup costs (tax, dis-synergies, one-time expenses) worth 8% of value is a marginal case. The strongest breakup candidates pair a discount above 20% with an available Section 355 structure.

Breakups Reveal Value; They Do Not Create It

A separation does not fix a weak business. A slow-growing, thin-margin segment that earns a low multiple inside the group will earn the same low multiple on its own. The breakup thesis requires a re-rating catalyst: at least one segment that the conglomerate structure is hiding from the investors who would pay properly for it. GE's three-way split into aerospace, healthcare, and energy businesses is the modern reference case of exactly that dynamic, with each piece re-rating once sector-focused investors could own it cleanly. Kellogg's split into Kellanova and WK Kellogg cuts the other way as a caution: the value was ultimately realized because Mars paid roughly $30 billion for Kellanova and Ferrero paid $3.1 billion for WK Kellogg, acquisition outcomes that were far from guaranteed on separation day.

The same caution applies to the SOTP output itself. A $25 gap between a $100 SOTP and a $75 stock price is not $25 of guaranteed upside. The discount exists for real reasons, and closing it demands action (spin-offs, sales, activist pressure) that takes years and carries execution risk and transaction costs.

SOTP Beyond Breakups

The same analysis anchors work that has nothing to do with splitting a company. In acquisition analysis, a buyer SOTPs a diversified target to see which segments justify the premium and which are non-core; in large-cap M&A it is common to announce divestitures of non-core segments alongside the deal, using proceeds to offset the price and cut pro forma leverage. SOTP also appears in fairness opinions on conglomerate transactions, alongside the standard methodologies, to show the price treats every segment fairly. In a pitchbook, the analysis typically lands as a waterfall from segment EVs through corporate adjustments to equity value per share, backed by mini-comps pages per segment and a direct SOTP-versus-share-price comparison.

Stub Value Analysis

Stub value analysis is a special case of SOTP where the market has already priced one of the parts. When a parent owns a stake in a publicly traded subsidiary, that stake has an observable market value; subtracting it from the parent's market cap reveals what the market implicitly pays for everything else the parent owns.

Suppose a parent carries a $30 billion market cap and owns 40% of a listed subsidiary worth $80 billion. The stake alone is worth $32 billion, so the stub is $30B minus $32B: negative $2 billion. In effect the market assigns a below-zero value to everything the parent itself operates, businesses that may be producing billions in revenue and profit.

Why Negative Stubs Persist

An obviously irrational price should be arbitraged away. Stubs persist because real barriers block the correction:

• •Tax friction. Monetizing the stake crystallizes capital gains: a $20 billion unrealized gain can imply $4+ billion of tax, a permanent cost that partially justifies the discount.
• •Short-selling constraints. The textbook arbitrage, long the parent and short the subsidiary, needs subsidiary borrow in size; borrow costs, scarce shares, and regulatory limits can make it impractical.
• •Management entrenchment. Distributing the stake shrinks the CEO's empire, so management resists moves that would unlock value for shareholders.
• •Contractual and regulatory restrictions. JV agreements, lock-ups, and approval requirements can freeze the stake for years.

The nature of the barrier drives the advisory answer. A tax barrier is a permanent cost, so part of the discount is rational and stays. A management barrier is addressable through activist pressure, so the discount is actionable. The banker's job is to size the discount and map the realistic path to eliminating it, not just to observe it.

Who Uses Stub Analysis

Activist investors screen for low or negative stubs as data-driven entry points: distribute the subsidiary shares, spin it off, or sell the stake and return proceeds. Bankers advise holding companies on the same menu, with a tax-free Section 355 distribution as the preferred route when it qualifies. Event-driven funds buy stubs that come with a catalyst and a timeline: a live activist campaign, a pending spin-off, an expiring lock-up.

The classic anchor is 3Com and Palm in 2000, where Palm's post-IPO market cap implied a negative value for 3Com's multi-billion revenue networking business; short constraints on Palm let the anomaly persist for months until 3Com distributed its remaining Palm shares. Vivendi's 2024 shareholder-approved split into four separately listed entities is the modern example, after years in which the holding structure priced Canal+, Havas, and the publishing assets below their independent worth.

One warning carries over from breakup analysis: a cheap stub is not automatically a buy. The core operations may be genuinely broken, separation may carry a prohibitive tax bill, or the holding structure may earn its keep through real benefits like shared services or access to the subsidiary's cash flow. Valuing the stub's underlying operations independently, by DCF or comps, is the mandatory next step before calling it mispriced.

Net Asset Value and Asset-Based Valuation

Nearly all banking valuation is earnings-based: multiples and DCF both price a company off the cash flow it generates. Asset-based valuation prices the company off what it owns instead:

For most operating companies the earnings-based answer is higher, because a going concern earns returns on top of its asset base: a software company's worth lives in customer relationships, code, and growth, not in its servers. But for a defined set of companies the assets ARE the business, and net asset value (NAV) is the correct primary methodology.

Where NAV Leads

• •REITs and real estate companies. Each property is valued individually, via cap rates applied to net operating income or property-level DCFs; NAV is the sum of property values minus debt and corporate obligations. The stock's premium or discount to NAV per share (P/NAV) is the standard relative metric.
• •Mining and natural resources. Each mine gets a life-of-mine NPV built from production plans, commodity price assumptions, and operating costs. Company NAV aggregates the mine-level NPVs together with development-stage projects and undeveloped reserves, then deducts net debt and the corporate cost center; P/NAV is again the primary multiple.
• •Holding companies and investment vehicles. Value equals the fair market value of the portfolio minus liabilities: observable for listed holdings, estimated by comps or DCF for private ones.
• •Distressed and liquidation scenarios. Liquidation analysis values each asset class at distressed recovery rates to establish the absolute floor for creditors; the mechanics belong to distressed valuation, but the logic is pure NAV.

The Floor, Not the Ceiling

For a healthy operating company, NAV functions as a valuation floor. A manufacturer holding $500 million of PP&E at fair value that generates $100 million of EBITDA earns a 20% return on its assets; at 10x EV/EBITDA it is worth $1 billion, double its asset value. The $500 million gap is the value of competitive position, customer relationships, workforce, and know-how, none of which sit on the balance sheet.

That framing makes the reverse signal diagnostic. Picture a business whose DCF or comps value comes to $400 million while its assets, marked to fair value net of liabilities, total $600 million: it may be worth more dead than alive, because the assets produce less as a going concern than they would fetch redeployed or sold. That comparison is the analytical foundation of the restructuring decision, reorganize and try to lift earnings, or liquidate and capture the asset value, and it also flags activist, breakup, and LBO opportunities where new owners could close the $200 million gap.

Where NAV Fails, and the Book Value Trap

Asset-based valuation is meaningless for asset-light businesses whose value lives in intangibles the balance sheet never records: a SaaS company can show $50 million of total assets against a $5 billion market cap; professional services firms are worth their people and client relationships, both of which can walk; consumer brands like Coca-Cola or Nike are worth multiples of their physical assets. Earnings- and revenue-based frameworks are correct there.

Equally important: NAV means fair market value, never book value. Book value records historical cost net of depreciation, and the two diverge violently over time; a building bought for $10 million in 1990 might sit on the books at $2 million while its market value is $30 million. A proper NAV revalues every material asset to current fair market value. Quoting book value as a valuation anchor for an operating company signals a basic misunderstanding of the method.

Dividend Discount Models

No valuation framework is older or theoretically cleaner than the dividend discount model: a stock is worth the present value of every dividend it will ever pay. The unlevered DCF dominates day-to-day banking work, but the DDM is the preferred intrinsic method for financial institutions and regulated utilities, where dividends are the equity cash flow that can actually be forecast and the unlevered framework breaks down.

The Gordon Growth Model

Myron Gordon and Eli Shapiro published the single-stage model in 1956. It treats the dividend as growing at one constant rate forever:

In the formula:

• •$V_0$ = equity value per share today
• •$D_1$ = next year's expected dividend, equal to $D_0 \times (1 + g)$
• •$r_e$ = cost of equity, from CAPM
• •$g$ = constant perpetual dividend growth rate

The formula is structurally identical to the perpetuity growth terminal value in a DCF and inherits the same constraints: $g$ must be below $r_e$ (or the value turns negative or infinite), and $g$ should not exceed long-run GDP growth of 2-3%, since no company can grow its dividend faster than the economy forever. As a numeric anchor:

A bank expected to pay $3.00 next year, with a 10% cost of equity and 4% perpetual growth, is worth $50 per share. The single-stage model earns its keep only under strict conditions: a long, stable dividend history; consistent growth (3-5% annually for banks, 4-6% for well-run utilities); a stable payout ratio (60-75% of earnings for most dividend payers in these sectors); and no expected acceleration or deceleration. For a company whose growth will change phase, the constant-growth assumption fails and so does the model.

The Two-Stage DDM

The two-stage model splits the future into a high-growth phase and a stable phase: an initial stretch of $n$ years at an elevated rate $g_1$, followed by a sustainable rate $g_2$ in perpetuity, with the terminal piece valued by the Gordon formula and discounted back:

It suits a company temporarily growing above its long-run rate: a bank aggressively expanding its loan book, or a utility recovering earnings after a rate case reset. A worked pass with a $2.00 current dividend growing 8% for three years, then 3% in perpetuity, at an 11% cost of equity:

1. 1.Stage 1 dividends: $2.16, then $2.33, then $2.52; discounted at 11% these are worth about $1.95, $1.89, and $1.84, summing to $5.68.
2. 2.Terminal value: the year-four dividend is roughly $2.60, so TV equals $2.60 / 0.08 = $32.50; discounted back three years (divide by 1.3676) gives $23.76.
3. 3.Intrinsic value: $5.68 + $23.76 = $29.44 per share.

The terminal piece is about 80% of the answer, the same dominance pattern the perpetuity term shows in a standard DCF.

The Three-Stage DDM and Bank Valuation

The three-stage model inserts a transition phase in which growth glides from the elevated rate down to the mature rate, rather than dropping in one step. This is the workhorse variant for bank valuation, where the stages map naturally:

1. 1.Explicit projections for 3-5 years, built from the bank's forecast earnings, its capital plan, and its payout target.
2. 2.A convergence phase of 5-10 years in which ROE fades toward the cost of equity as excess returns are competed away.
3. 3.A terminal stage growing at long-run GDP, valued with the Gordon formula.

A Levered Model with No EV Bridge

The DDM discounts an equity cash flow (dividends) at the cost of equity, not WACC, so it produces equity value directly. There is no enterprise value and no bridge. That is precisely why it fits banks: deposits and borrowings fund the lending book, so debt is an operating input rather than financing, and separating operating from financing cash flows, the first move of an unlevered DCF, is impossible.

The input that converts an earnings forecast into a dividend forecast is the dividend payout ratio, dividends per share over earnings per share: Dividends = Earnings x Payout Ratio. For banks the ratio is bounded from above by regulatory capital requirements; CET1 minimums and Federal Reserve stress tests (CCAR and DFAST) cap distributable capital, and banks typically target 35-50% payout in dividends alone or 60-80% counting buybacks. That constraint is also why the DDM works: banks cannot hoard earnings indefinitely, so distributions really are the shareholder's return. Regulated utilities fit for the parallel reason that allowed returns on a regulated asset base make dividend growth predictable, payout ratios sit steadily at 60-75%, and the income-oriented investor base already prices the stock in dividend terms.

When Not to Use It, and How to Sanity-Check It

The DDM is the wrong tool for growth companies (no dividends, often no earnings), cyclicals (dividend cuts in downturns break the growth assumption), companies with irregular payout histories (value accrues through capital appreciation, so use a DCF), and companies in transformational change (dividends may be suspended or reset).

Where it does apply, reverse-engineer the market's assumption as a check. Rearranging Gordon Growth gives the implied growth rate $g = r_e - D_1 / P_0$. A bank at $50 paying a $2.00 dividend with a 10% cost of equity implies growth of 10% minus the 4% dividend yield, so 6%, above long-run GDP. Either the market genuinely expects above-economy growth, or the cost of equity estimate is too high; and if the bank has historically grown earnings at 4%, the stock looks rich on DDM assumptions.

Respect the model's sensitivity when presenting output. Because $g$ sits in the denominator, a half-point change can move the value 15-25%, and the effect amplifies as $r_e$ and $g$ converge: at a 10% cost of equity, lifting growth from 3% to 4% shrinks the denominator from 7% to 6%, a one-seventh drop that lifts the implied value by roughly 17%.

Choosing Among These Methods

The common thread is diagnostic. Reach for SOTP and breakup analysis when the segments of a company belong to different valuation paradigms; reach for stub value analysis when the market has already priced one piece and left the remainder implicitly cheap or free; reach for NAV when value resides in the assets themselves, or as a floor and a dead-or-alive test for a struggling operator; reach for the DDM when dividends are the only projectable equity cash flow and enterprise value is not meaningful.

None of them replaces the core toolkit. Each redeploys comps, DCF, and cost-of-equity logic at a different unit of analysis: the segment, the asset, the dividend stream. The interview skill is spotting the moment the consolidated, going-concern, unlevered frame stops describing the company in front of you, and saying which of these methods takes over and why.