Infrastructure DCF

Building a valuation for a real asset whose cash flow has a known end date follows a fixed sequence: (1) establish the remaining concession or contract life, N years; (2) forecast revenue for each year, traffic or tariff or contracted price times volume; (3) forecast operating expense; (4) deduct a major-maintenance or lifecycle-capex reserve (a real cash cost distinct from ordinary opex); (5) build the debt schedule, interest on the outstanding balance plus scheduled principal amortization, for every year; (6) compute distributable cash flow to equity as EBITDA minus the maintenance reserve minus interest minus principal; (7) discount the resulting stream at a rate appropriate to the asset's risk tier (see the infra/real-asset risk spectrum); and (8) set the terminal value to zero (or to a contractually stated handback/salvage value, if the concession agreement specifies one) at year N.

Why is the terminal value zero in an infrastructure DCF?

Read step (8) again, because it is the single most important adjustment this module makes to the machinery you already own. M3.05 built the going-concern terminal-value discipline for a business that keeps existing after your explicit forecast window ends: terminal growth capped at the risk-free rate of the same stack the discount rate was built from, which is 4.5% on the cleaned INR stack, plus a steady-state return-on-invested-capital test. A concession does not keep existing in your hands: the asset, or at least your claim on its cash flows, reverts to the government grantor (or the contract simply expires) on a date written into the concession agreement itself. Setting the terminal value to zero here is not a conservative simplifying assumption you are choosing to make, it is the contract, and treating it as anything else (a Gordon-growth perpetuity, a REIT-style exit multiple) invents value that legally does not exist. Worked Example 3 (an Indian toll-road concession) builds this fully; Worked Example 4 (a US wind asset) extends it to a two-stage case where a contracted period gives way to an uncertain "merchant tail" that is genuinely valuable but genuinely riskier, and therefore, correctly, discounted at a different rate rather than dismissed to zero out of caution.

Modern Data Analysis note. Every schedule in this section is a short, mechanical loop once you have the logic straight: a for loop over the debt schedule (interest on the opening balance, a fixed or sculpted principal repayment, a running closing balance), one line applying a growth rate to revenue, and a one-line discounting sum at the end. Writing it once as a small reusable function (concession_dcf(revenue0, growth, opex_pct, maintenance, debt0, rate, life, discount_rate)) that you can re-call for every deal, rather than rebuilding a fifteen-row spreadsheet by hand each time, is exactly the habit the quant-code region's Python/pandas track will drill into a full toolkit, including a proper root-solver for the IRR instead of trusting a spreadsheet's black-box IRR() function (Worked Examples 3 and 6 both hand-check an IRR by bisection rather than assuming a function got it right). This module teaches you what to compute and why it must be structured this way; that region teaches you never to hand-build the same table twice.