valuation
Sum-of-the-parts (SOTP) valuation values each business segment with a method and peer set that fit that segment, then adds them up and subtracts net debt, NCI and holding-company costs. It is the tool for conglomerates and mixed models a single multiple cannot describe.
A single EV/EBITDA on a group that mixes a software division, a low-margin distribution arm, and a property company is not conservative—it is incoherent. Sum-of-the-parts valuation (SOTP, break-up value when used as a floor) assigns each part a method: usually comparable companies or a segment DCF, sometimes a NAV for assets.
Equity value = Σ segment EVs − net debt − holdco overhead (NPV) − NCI + surplus non-operating assets
IFRS 8 segment disclosures are a starting map, not a guarantee that the segments are how a buyer would carve the group.
Central costs that do not disappear in a split must be valued (capitalise a run-rate holdco EBITDA loss, or NPV the cost). Tax at holdco, pensions, and listed-related overhead belong here. A "conglomerate discount" versus the SOTP is an observation about listing and complexity, not an input you should bake in twice. In a private sale of the whole group, the discount may be smaller than in a sleepy listing; in a break-up, execution costs and tax leakage must be subtracted explicitly.
Atlas Retail Group is invented, illustration only.
| Segment | Method | EV (€m) |
|---|---|---|
| Grocery banners | 6.5x EBITDA €40m | 260 |
| Specialty e-commerce | 1.8x revenue €55m (earnings still noisy) | 99 |
| Property (owned stores) | Cap rate 6.0% on €18m NOI | 300 |
| Less: NPV of holdco costs | (45) | |
| Group operating EV | 614 | |
| Net debt | (190) | |
| NCI in a banner JV | (22) | |
| Equity value | 402 |
A naive 8x on blended EBITDA €48m would have produced EV €384m and missed that property and e-commerce do not share a multiple. The SOTP can be higher or lower than a blended multiple; the point is the map.
Activists, break-up bankers, and corporate development teams live in SOTP. Buyers of one division use it to avoid paying for parts they will not get. Sellers use it to justify retaining a jewel or to set a floor versus a lowball offer on the whole. Segment DCFs need allocated capex and working capital, not just allocated EBITDA.
Allocations of central cost and capital are judgement. Markets may never pay the SOTP if the parts cannot be separated cleanly. Thin segment comps are still thin. SOTP precision is usually lower than a single-business DCF, not higher, because you have stacked several uncertain models.
Dazur reaches for SOTP whenever a CIM's blended multiple is doing too much work—typical in Iberian groups that mix operating companies with property and a small digital arm. We would rather value three honest pieces and a holdco cost than one glamorous average. Buyers of a part, and sellers of the whole, should see the same map.