Terminal value

A steady state after year five

TANGE starts the continuing period from NOPAT in the fifth explicit year. That NOPAT grows once at the perpetual growth rate. The model then deducts the investment required to support that growth before capitalising the resulting terminal free cash flow.

The explicit plan and the terminal period need not have the same return on invested capital. The plan starts from the company’s own profitability, while the terminal period uses the sector’s steady-state return on capital. The difference is an explicit convergence assumption, not a reconciliation error.

Growth requires reinvestment

The implied reinvestment rate is perpetual growth divided by steady-state return on invested capital. This relationship makes the operating assumption explicit: a business cannot keep increasing NOPAT without committing capital. A perpetuity that grows but reinvests nothing would create value from an economic resource that the model never funds.

Implied reinvestment rate = g∞ / ROIC∞

Terminal FCFF = NOPAT(5) × (1 + g∞) × (1 − g∞ / ROIC∞)

The model requires ROIC∞ to exceed g∞. Otherwise the steady state would consume all—or more than all—of operating profit merely to sustain growth.

Capitalisation and discounting

The terminal free cash flow is capitalised with the spread between WACC and perpetual growth. The value is measured at the end of year five and discounted back for 4.5 years, the same mid-year convention used for the explicit flows.

Terminal value at year 5 = terminal FCFF / (WACC − g∞)

Present value of terminal value = terminal value / (1 + WACC)^5

Why its share is shown

The terminal share is the present value of the terminal value divided by enterprise value. It is reported prominently because a high share means that much of the estimate depends on assumptions beyond the explicit plan. It is a diagnostic of where the result comes from, not a separate valuation adjustment.

One value, and a measured sensitivity around it

Earlier versions of this tool showed a low–central–high band, built by moving the WACC half a point either way and the fifth-year margin one point either way. The band was withdrawn. Those two movements were a convention, not a measurement: they were neither the two parameters that matter most in every case, nor a probability statement about the result, yet a printed interval reads as one. What replaces it is the sensitivity table on the result surface, which moves each of the nine parameters in turn by the same fraction of its own slider range and reports how far the value travels. It says which assumptions the estimate rests on, which is the question the band was being asked to answer.

The ceiling on perpetual growth, and why it is the risk-free rate

Perpetual growth is capped at the risk-free rate. When the cap binds, the tool says so on screen: the estimate uses the capped figure, not the one requested.

The reason is arithmetic before it is methodological. A perpetuity growing for ever faster than the economy it lives in eventually exceeds that economy, so the growth rate of a going concern in perpetuity is bounded by the growth of the economy around it. The long-term risk-free rate is the usual stand-in for long-run nominal growth: it is already in the model, with its own source and its own observation date, so the ceiling introduces no new number to defend.

III.1.38 is the anchor. Among the things the appropriate terminal estimate should rest on, the ED lists “a growth rate consistent with the fundamentals”. A rate above the growth of the economy is not consistent with any fundamental: it is a rate the business could sustain only by absorbing an ever larger share of everything, for ever. The ceiling is the mechanical part of that requirement — the part a form can enforce. The rest of it, the cyclicality of the business and the erosion of competitive advantage, remains judgement, and remains outside what this tool does.

The ceiling replaces the 2% cap that the previous version declared in its sector data and that fell with the operating block. It is not the same number and does not pretend to be: it is a rate with a source, which moves when the rate moves, rather than a constant nobody could re-derive.

Against III.1.38: this is the mechanical application the ED tells you to avoid

Reference is to the Exposure Draft — Principi Italiani di Valutazione 2026, Fondazione OIV, December 2025. It is a draft in consultation, not a standard in force.

III.1.38 requires the valuer to avoid mechanical applications of the DCF formula in estimating terminal value, and names as its example “the estimate of the long-term result flow as the product of the last plan year’s result flow and the annual growth rate forecast from the period after the explicit forecast”. That is exactly what this tool does. The terminal flow here is year-five NOPAT grown by g and reduced by the reinvestment g/R. There is no way to read the two in agreement.

What the same paragraph asks for instead is a sustainable long-run result flow, arrived at by considering the cyclicality of the business, a growth rate consistent with the fundamentals, the investment needed to sustain it, the possible erosion of the sources of competitive advantage, and the risk profile of that sustainable income. Every one of those is a judgement about a specific business. None of them can be read off a balance sheet by a form. This is not a defect that better code would remove, and it is not a defect of this tool in particular: it is inherent to any automated calculation. Every tool that computes a terminal value from a form is in the same position, because the principle reserves that step to reasoned judgement about a specific business, and a form has no judgement to reserve it to. The difference between one such tool and another is not whether the limit is present but whether it is declared.

What the tool does do is make the mechanism visible rather than hide it. The steady-state return is derived from the plan instead of being taken from a table, it is declared as derived, forced or capped, and the three returns — base year, year five, steady state — are shown side by side so an implausible profitability is visible at once. That mitigates the mechanical application; it does not satisfy III.1.38.

The cross-check the principle recommends is not available here either. III.1.38 adds that a comparison with an estimate based on market multiples of suitably chosen comparable companies can provide a useful control. This tool has removed the multiples branch: the sector EV/EBITDA was null for every sector, so the comparison was dead code, and the methodological objection stands anyway — listed-company multiples carry a liquidity premium that overstates disposals of unlisted businesses. The consequence is that the terminal estimate here runs without the control the principle suggests for it. Removing a comparison that never worked was right; it does not restore the control, and whoever uses this number should obtain that comparison elsewhere.