Case study: General Electric, the conglomerate

A conglomerate must be valued as its parts, sum-of-the-parts, because one blended multiple cannot price unrelated businesses, and GE's 2021-24 breakup largely vindicated that lens: the separated companies were together worth far more than the whole had traded for.

For two decades GE was the most admired company in America and its CEO the era's management icon; within another two it had lost around three-quarters of its peak value, cut its dividend to a penny, and finally split itself into three. The full arc is the best education that exists in analyzing conglomerates, because every classic conglomerate pathology appeared in sequence, in public, with filings to read.

The short version

A conglomerate must be analyzed as its parts (sum-of-the-parts: each segment valued by its own industry's method, minus corporate costs and debt) because one blended multiple cannot price three unrelated businesses. GE added the two great complications: a finance subsidiary that made consolidated statements nearly unreadable and hid enormous tail risk, and a legendary earnings-smoothness that was itself the red flag. The 2021-24 breakup largely vindicated the SOTP lens: the separated aerospace, healthcare and energy companies were together worth far more than the conglomerate had traded for. Discounts to SOTP are real, but they need a catalyst, and they can persist for a decade first.

Why this case

A conglomerate is several unrelated businesses under one ticker: at its 2000s peak, GE was jet engines, gas turbines, MRI machines, plastics, television (NBC), light bulbs, locomotives, and one of the world's largest finance companies. No single multiple, margin benchmark, or peer set applies to that collection, which is precisely the analytical problem: the consolidated statements ADD together businesses whose economics have nothing in common, and the analyst's first job is to take them apart again.

The arc, 1981-2024

Period	Event	The signal available at the time
1981-2001	The Welch era: relentless EPS growth, ~100 quarters of 'making the number'; peak value ~$600B in 2000, the world's most valuable company	The smoothness itself: industrial + finance earnings do not naturally arrive that evenly
2001-2007	Immelt inherits the peak; GE Capital grows to roughly half of profits: insurance, mortgages, commercial paper funding	Segment notes showed an industrial company financing long assets with short paper
2008-2009	GE Capital nearly fails in the funding freeze; dividend cut ~68%; Buffett rescue capital; AAA rating lost	The finance sub was the company; the industrial moat could not save the balance sheet
2015-2017	GE Capital wound down (the 'good' decision); Alstom power acquisition at the top of the gas-turbine cycle (the bad one); buybacks near peak prices	Capital allocation score-keeping: selling low, buying high, returning cash it did not have
2017-2018	The reckoning: dividend cut twice (to $0.01), a ~$6B insurance reserve shortfall surfaces from long-closed books, SEC accounting investigations, removed from the Dow after 111 years	Legacy liabilities from a finance book outlive the earnings they once smoothed
2021-2024	The breakup: GE HealthCare (2023), GE Vernova (energy, 2024) spun off; GE Aerospace remains	The SOTP made real: the three parts together promptly exceeded the whole's prior value

Why the consolidated numbers resisted analysis

A bank stapled to an industrial. GE Capital's balance sheet (hundreds of billions of financial assets funded heavily in short-term commercial paper) consolidated into the industrial statements, making every classic ratio (leverage, coverage, working capital) meaningless without full segment separation. Rule: any conglomerate with a finance arm must be split into FinCo and IndustrialCo before a single ratio is computed, each analyzed with its own industry's tools (the bank methods from Other Intrinsic Methods for the FinCo).
Earnings smoothing as a feature. The famous quarter-after-quarter precision was achievable because a finance book contains timing levers: gains on sales, reserve releases, tax items. Analysts who asked "what would earnings look like WITHOUT the finance offsets?" found lumpier, lower-quality industrials underneath. Smoothness in a business that should be lumpy is not comfort; it is a question.
Cash flow told it earlier. Industrial free cash flow persistently lagged reported industrial earnings in the 2010s: long-term service accounting (recognizing profit on multi-decade engine-service contracts ahead of cash) flattered the income statement. The statements guide's first red flag, at $100B scale.

The SOTP, done properly

A late-2018-style sum-of-the-parts (illustrative mechanics)

Aviation:    ~$4.5B segment EBITDA x 12-14x (aero peers)   = $55-65B
Healthcare:  ~$3.5B EBITDA x 13-15x (medtech peers)        = $45-55B
Power:       depressed EBITDA; value on recovery scenarios = $10-25B
Renewables, other industrial                                = $5-10B
GE Capital:  book value LESS runoff/insurance risk         = $0 to -$10B
- corporate costs capitalized (~$2B/yr at ~8x)             = -$15B
- net industrial debt & pension deficit                    = -$50B
=> equity range                                            ≈ $50-120B

...against a market cap that touched ~$65B. The lesson is not the
midpoint; it is the RANGE: the answer hinged on Power's recovery and
on how much of GE Capital's tail you believed. SOTP done honestly
exposes WHERE the disagreement lives; done dishonestly it launders
optimism through six small overstatements that compound.

Each part gets its own method: aero on EBITDA multiples against its duopoly peers, healthcare against medtech, power on normalized mid-cycle economics (a cyclical inside the conglomerate!), the finance book on adjusted book value.
Corporate costs are a real negative asset: capitalize the unallocated center like a perpetual expense; enthusiasts habitually forget it.
Pensions and legacy liabilities belong in the bridge: GE's multi-billion pension deficit and insurance reserves were equity claims senior to shareholders, exactly like debt.
Then demand the catalyst: a discount without a mechanism (activist, breakup, management change) can persist indefinitely; GE traded below plausible SOTP for YEARS before the split unlocked it. The precedent-transactions guide's rule applies: an event-priced value needs an event probability.

The GE Capital lesson, stated as a rule

The finance subsidiary trap

A finance arm lets an industrial borrow cheaply against its industrial reputation, lever the proceeds into financial assets, and report the spread as smooth "industrial-quality" earnings, until funding markets close, when it turns out the company's survival depends on rolling commercial paper like any bank, but without a bank's deposits or regulator-forced capital. 2008 proved the funding fragility; 2018's insurance charge proved the LIABILITY tail: books closed decades earlier produced a ~$6B cash call. When any company's finance arm exceeds ~20-30% of earnings, analyze the whole enterprise as a leveraged financial with an industrial attachment, demand bank-style disclosure (funding mix, duration match, reserve adequacy), and price the opacity itself as a discount.

The breakup verdict

The 2021-24 split is the rare natural experiment: the SOTP thesis got tested with real market prices. Freed from the conglomerate, GE Aerospace commanded a premium aero multiple, HealthCare a medtech multiple, and Vernova, the power business everyone had written off, rerated dramatically as the grid/electrification cycle turned: the three together soon traded far above the combined company's pre-split value. The vindication carries its own nuances: part of the gain was the businesses IMPROVING under focused management and a cycle turning (Vernova), not pure discount-closing, and that ambiguity is permanent in breakup investing: the discount, the focus dividend, and the cycle arrive together, and the analyst's attribution is always partly guesswork. What is not ambiguous: the blended-multiple view had underpriced the collection for years, exactly as the SOTP lens claimed.

The transferable lessons for any conglomerate

Never analyze the blend. Segment notes first; every part valued by its own industry's method; corporate costs, pensions and legacy liabilities in the bridge at full weight.
Treat suspicious smoothness as a finding. Ask what mechanism produces it and what it hides; check earnings against segment-level cash.
Ring-fence any finance arm. FinCo/IndustrialCo split, bank tools for the FinCo, and a discount for opacity, not a premium for "diversification."
Score capital allocation across a full cycle. GE bought high (Alstom, buybacks at the top) and sold low, in public, for years: the record was in the filings before it was in the stock.
The discount needs a catalyst, and patience needs a budget. SOTP gaps are real and can persist a decade; position sizing must survive the waiting, and the thesis doc's time-stop applies.
Focus has a value of its own. The breakup's gains came partly from management attention, incentives and investor clarity, a recurring finding across spin-offs, and the reason the spin-off calendar is a permanent hunting ground.

Glossary for this guide

Sum-of-the-parts: Valuing a multi-business company piece by piece, each segment by its own industry's method, then summing and netting corporate costs and debt. The tool for conglomerates and the arithmetic behind every break-up thesis.
Conglomerate discount: The recurring finding that a diversified company trades below the sum of its parts. Either the analyst's parts are overpriced or the structure itself destroys value; activists exist to argue the second.
Book value: Shareholders' equity as the balance sheet states it: assets minus liabilities. Meaningful where assets are financial or tangible (banks, insurers), nearly meaningless where the real assets are brands and code.
Seniority: The bankruptcy queue: secured lenders, unsecured bondholders, subordinated debt, preferred, then equity last. The same company's different claims can deserve opposite verdicts because they stand in different places.
Mid-cycle earnings: A cyclical company's earnings averaged across the cycle rather than taken at the current peak or trough. Valuing cyclicals on spot earnings buys high and sells low by construction.
Catalyst: The event expected to force the market to reprice toward your view: earnings, a spin-off, a regulatory decision, a contract. Cheap without a catalyst can stay cheap for years; the catalyst is the thesis's clock.
Capital allocation: What management does with the cash the business throws off: reinvest, acquire, repay debt, buy back stock, pay dividends. Five years of these choices, priced against what they returned, is a CEO's real report card.
Non-GAAP measures: Company-defined numbers like adjusted EBITDA that exclude items the company chooses to exclude. Filings must reconcile them to the audited figure; the reconciliation, and especially its size, is the informative part.
Footnotes: The notes attached to audited financial statements: debt maturities, leases, segments, pensions, litigation, accounting policies. The fine print that professionals read first, because it is where inconvenient detail is required to live.
Maturity wall: A cluster of debt coming due in a narrow window. Refinancing risk has a calendar, published in the debt footnote, and a fine company with everything due in a shut market is a default candidate.
Float: Premiums an insurer holds between collecting them and paying claims: other people's money available to invest. Cheap, durable float compounding in good hands is the engine Berkshire Hathaway was built on.