A pattern recurs at investment committee tables with enough regularity to be treated as structural rather than incidental. The team presents a model in which a downside case already sits beside the base case, the production assumption has been held within a conservative band, operating expenditure escalates above inflation, and the revenue line carries a deduction for counterparty default. When the presentation ends, the first comment from the table rarely contests any of those assumptions; it turns instead to the discount rate, and someone observes that, given the geography, the technology, and the counterparty in question, the rate ought to be taken up somewhat. The rate is taken up, the project fails to clear the hurdle, the file closes. Asked afterwards which risk was priced, the room would name one; that risk, however, has been deducted twice, in two separate places, through two separate mechanisms.
The pattern is not peculiar to the investment committee. In the preparation of an EPC bid, the contractor embeds its own reserve inside unit rates, the owner's cost consultant layers an owner's contingency on top of the resulting number, the financing side then requires a separate contingency facility against the same construction exposure, and a concealed float is finally inserted into the programme. The same stacking appears around equipment failure, where machinery breakdown insurance is purchased, a portion of the manufacturer's warranty is withheld from payment, a maintenance reserve account is funded, and an availability haircut is applied in the model in addition. Each of these layers is individually defensible, and each was adopted by a competent party for a stated reason; the aggregate is visible nowhere, because no one has ever set the four items side by side on a single page.
The mechanism at work is the double counting of risk: the same uncertainty penalised a second time, whether across the cash flow and the discount rate or across several distinct instruments of prudence. Its origin lies in the dual role the discount rate performs in institutional practice. In theory the rate compensates for non-diversifiable systematic exposure and for the structure of the capital that funds the asset; in practice it functions as the vessel into which every discomfort that cannot be modelled, cannot be named, or would take too long to argue is quietly poured. Where country exposure is added in one review round, technology exposure in the next, and counterparty quality in a third, the rate has ceased to be a cost of capital and has become the collective unease of the institution compressed into a single figure.
Under certain conditions the tendency is entirely functional and should not be read as an error. Where information is thin, where the data set is narrow, or where the decision window is short, a premium placed on the rate is an inexpensive shortcut that introduces prudence into the system without bearing the cost of modelling each exposure separately. To this must be added the asymmetry facing whoever conducts the review: the institutional cost of approving a bad project is visible and personally attributable, whereas the cost of declining a good one is never measured and is invoiced to no one. Adding a further buffer is individually rational under that asymmetry. The difficulty lies not in the shortcut itself, but in its persistence once the risk in question has already been explicitly modelled elsewhere.
The distinction that resolves this is short and operable. Project-specific, diversifiable exposures — construction delay, production variance, the default of a single counterparty, the extension of permitting timelines — enter expected cash flow through probability-weighted scenarios; they do not belong in the rate, because at portfolio level they partially extinguish one another. Systematic exposure, meaning the commodity cycle, the level of interest rates, the elasticity of demand and comparable exposures from which diversification offers no escape, belongs in the rate. Where that boundary is left unwritten, the line between the base case and the prudent case is drawn differently by every team, the base case silently becomes a downside case, and the rate continues to load a premium on top of a downside case that has already absorbed the deduction.
The first layer of institutional cost is arithmetic, and it compounds with tenor. A deduction applied within the cash flow flows through each year in linear fashion, whereas a premium added to the rate penalises distant years exponentially; double counting therefore leaves short-payback work built on existing plant almost untouched, while systematically eliminating long-tailed assets that generate no cash in their early years and carry their value across the operating period. The outcome is not a strategic choice that anyone deliberately made. It is the drift of a portfolio in a direction no one selected — toward short-dated, incremental, low-return work. What determines the growth profile of an institution is frequently not strategic intent but the accumulated sum of premia added to the discount rate over the years and never subsequently withdrawn.
The second layer surfaces in the financing structure. The sizing case a lender applies to debt capacity is conservative by construction; the DSCR threshold is tested against depressed production and elevated expenditure, and a debt service reserve account together with cash trap mechanics is layered on top. When the sponsor model adopts that structure wholesale and then applies its own prudent production assumption and an elevated equity hurdle above it, the same downside exposure has been priced three times: once in the debt quantum, once in the reserve account, and once in the required return on equity. The practical consequence of that triple load is an equity requirement larger than the asset genuinely warrants, with the sponsor committing capital to fund a scenario that will not occur.
The third layer sits at the negotiating table and is the most expensive of the three. In an acquisition process the buyer first applies a haircut to the seller's projections, then takes the discount rate up on the grounds of the target's scale and its dependence on the founder, and then requires an earn-out structure together with expanded representations and warranties against the same uncertainty. Viewed from the seller's side, one risk has been priced in four separate places, and the resulting price gap arises not because the parties disagree about the magnitude of the exposure but because neither side has kept a tally of how many times it has been counted. A share of transactions that fail to close reflect not a genuine dispute but an accounting error.
This tendency cannot be managed through individual awareness, because each step of the double count is taken by a different person, in a different meeting, for an entirely reasonable stated reason. What can be managed is the architecture, and the core of that architecture is a single rule: each risk is priced in one instrument only. The mechanism carrying that rule is a column added to the risk register which makes the argument visible — against each line, the instrument in which that risk has been priced is recorded, whether in a scenario weighting, a discount premium, a reserve account, an insurance policy, a liquidated damages cap, or a warranty in the sale agreement. Where two marks appear on the same line, the decision becomes which of the two instruments survives; retaining both is not a choice but an oversight.
A second component completes it: ownership of the discount rate is removed from the deal team. Where the rate is set and written down at portfolio level, by a body separate from the transaction and in decomposed form — risk-free return, systematic risk premium, capital structure component — the suggestion in a live negotiation that the rate be taken up somewhat automatically resolves into the question of which component is moving and on what evidence. The third component is the standardisation of scenario definitions: fixing across the institution that the base case represents an expected value rather than a prudent value closes the door through which double counting most frequently enters, since it is the silent migration of the base case toward a downside case that leaves room for the second deduction.
In the capital-intensive projects it manages, BEIREK consolidates these three components into a single risk allocation record and reopens that record at each gate decision preceding FID; the record binds every risk to one instrument and documents the withdrawal of the prior pricing whenever the instrument changes. On the financing side a reconciliation round is run that compares the lender's sizing case with the sponsor model's assumption set line by line, its sole purpose being to make visible whether a downside exposure already priced within the debt structure is being priced a second time in the equity hurdle. The same discipline extends to the contract line, where the LD cap, the performance guarantee, the insurance programme and the reserve accounts are aligned in a single table, and structures that address one event through three instruments are simplified before the redundancy is loaded into the bid price.
Prudence is measured not by how far an institution takes its numbers down, but by its ability to demonstrate where each risk has been priced; the first is a posture, the second a method. Looking back at an investment decision, the question worth asking is not whether the asset was valued conservatively enough, but how many separate line items absorbed a deduction for the same uncertainty — and because in most institutions the answer appears in no document, the record capable of producing that answer precedes the quality of the decision itself.