Sum-of-the-Parts Valuation: When the Conglomerate Discount Is Permanent

Sum-of-the-parts (SOTP) is the valuation method analysts reach for when a single multiple won't do — when a company is really three or four businesses stapled together, each with different growth, margins, and capital needs. Done well, it surfaces real mispricings. Done badly, it generates phantom upside that never closes.

The core question isn't "what are the parts worth?" It's "why doesn't the market already give them credit?" A discount that exists for a structural reason is not an opportunity. A discount that exists because of a fixable friction is.

How to build a defensible SOTP

The mechanics are straightforward. Split the company into segments that have economic independence — not just reporting segments, but units that could plausibly be sold or spun. Apply a peer multiple (EV/EBITDA, EV/Sales, or a DCF) to each, sum the enterprise values, subtract net debt and minority interests, add cash, and divide by shares.

Three places people cheat themselves:

Peer selection. A captive finance arm inside an industrial is not Visa. A media unit inside a telco is not Netflix. Use pure-play peers that match the segment's actual margin and capital profile, not its aspirational one.
Corporate overhead. Unallocated SG&A, pension obligations, and stranded costs don't vanish in a spin. Allocate them, or carve out a separate negative-value bucket for "HoldCo cost." A clean SOTP that ignores corporate overhead is just wrong.
Tax leakage. Selling a business with a low tax basis triggers a real bill. Spins can be tax-free under Section 355 if structured correctly, but sales rarely are. Haircut accordingly — often a material percentage of the gain.

A credible SOTP shows the bridge from gross asset value to equity value per share, line by line. If you can't reconcile it to the current market cap and explain the gap, you don't have a thesis — you have arithmetic.

When the conglomerate discount is real (and closeable)

A "real" discount is one tied to a friction that management can remove. The classic catalysts:

Hidden compounders. A high-growth segment buried inside a slow-growth parent gets valued on blended multiples. The market literally cannot see it until it's separated. This was the bull case on AT&T/WarnerMedia, on Dell/VMware, and on countless industrial spins.
Capital allocation conflicts. When cash from a mature cash cow funds a struggling sister business, both get penalized. Splitting them lets each pursue its natural capital structure — the cash cow levers up and returns capital, the growth business raises equity or reinvests.
Coverage and index mismatch. A media business covered by industrials analysts trades at industrial multiples. Post-spin, it gets re-covered by media analysts and re-rated.
Activist pressure with a credible plan. When a serious activist forces the question, the discount often compresses before the actual separation.

The common thread: there's a specific mechanism by which value transfers from the SOTP spreadsheet to the share price. "It should be worth more" is not a mechanism.

When the discount is permanent

Some discounts persist for decades because the underlying reason is structural. Recognize these and stop pretending:

Controlling shareholders. Dual-class structures, family control, or government stakes (think Samsung, LVMH, many European holdcos) mean minority shareholders cannot force a separation. The discount is the price of the governance.
Cross-subsidization that's load-bearing. Sometimes the "trapped" cash cow is actually paying for the customer relationships, distribution, or balance sheet that makes the other businesses viable. Berkshire's insurance float funding the equity portfolio is the cleanest example — the discount, if any, is the price of the structure that creates the alpha.
Tax basis lock-in. A business held for decades with near-zero basis cannot be sold without a punitive tax hit. The discount reflects after-tax separation value, which is the right number.
Diseconomies of de-merger. Shared technology platforms, joint procurement, regulatory licenses, or pension liabilities that can't be cleanly split. Stranded costs post-separation can exceed the multiple uplift.
Persistent value destruction at HoldCo. Some conglomerate parents have spent years buying and selling businesses with negative net IRR. The market is right to discount future capital allocation.

A quick test: has this discount existed for more than five years with no narrowing despite changes in management or market conditions? If yes, assume it's structural until proven otherwise.

A framework for sizing the opportunity

Before acting on a SOTP gap, work through four questions:

What's the catalyst? Spin announcement, activist campaign, CEO change, refinancing wall, or strategic review. Without one, you're paying for hope.
What's the timeline? Most spins take 12-24 months from announcement to completion. Discount your upside back at your hurdle rate.
What's the residual stub worth? After the attractive piece spins out, what's left? Often the answer is "a worse business than the consolidated entity" — because the cross-subsidy is gone.
What does management actually own? If insiders hold meaningful equity and have publicly committed to a review, the probability weighting shifts.

If three of four answers are weak, the SOTP gap is probably permanent.

What to watch next

Pull the 10-K segment footnote for any conglomerate you own and rebuild a rough SOTP using current peer multiples. Compare to enterprise value.
Check insider ownership and dual-class structure before assuming any discount will close. Governance trumps math.
Track activist 13D filings and proxy fights in your portfolio — these are the highest-probability catalysts for SOTP realization.
Re-underwrite the stub in any spinoff situation. The piece you're left holding often has worse economics than the headline math suggests.