A project typically arrives at an investment committee in two documents rather than one. The first carries the technical and commercial assumptions and is built on unlevered cash flow, producing a return figure that occupies a single line on the summary page. The second, prepared by the financing team, sets out the tranche structure, the margin, the tenor profile, and the balance of sources and uses. Even when both are discussed in the same sitting, they are not tied to the same arithmetic: the first is read as governing whether the project proceeds, the second as governing how an already-approved project will be funded. The number that determines ranking comes, almost invariably, from the first document. Yet several of the line items sitting in the second — arrangement fees, commitment charges, letter-of-credit costs, cash immobilised in a debt service reserve account, the cost of interest rate hedging — are economic consequences borne by the owner of the project, and they are frequently large enough to alter the decision itself.
The same separation shows up on the calendar. In most institutions preliminary approval precedes the term sheet; by the time financing terms crystallise, the decision has been taken, the capital allocated, the team assembled. When the incremental cost imposed by the credit terms finally surfaces, it is recorded not against project return but under closing costs, and the ranking is not recalculated, because the page on which the ranking rests is never reopened. This is not a calculation that someone disregards; it is a calculation that no role explicitly owns. The technical team treats financing terms as given data, the financing team treats project economics as given data, and the territory where those two sets of data intersect belongs to neither.
The pattern has a name — financing-side-effect neglect, the exclusion from project analysis of the value effects generated by the financing decision — and its mechanism runs as follows: the valuation treats the asset side of the project as a whole and the financing side as a neutral conduit through which capital passes without altering what is created. The assumption is rarely written down, but it is this: however capital is raised, the value produced is identical, and therefore the asset side alone suffices for comparing projects. The convenience this assumption delivers is genuine. It renders dozens of projects comparable on a single page, concentrates debate on the technical assumptions where the operating teams hold real expertise, and permits decisions to be taken while financing terms remain unsettled.
The conditions under which the shortcut remains functional are narrow, but they are real. Where every project in a portfolio is financed at similar leverage, through similar instruments, and under the same legal entity; where the group's tax position is stable and wide enough to absorb the shield in full; and where financing costs move within a narrow band relative to capital outlay, excluding financing effects will not disturb the ranking. Under those conditions the effect falls on each project in roughly the same proportion, so keeping it outside the calculation costs a level of precision without costing direction. Where the decision rests on ordering, and the ordering survives intact, the shortcut is effectively free.
The difficulty arises when conditions change and the shortcut does not. Where the same committee is presented, on one side, with a contracted long-dated asset capable of carrying high leverage under a project financing structure and, on the other, with an investment fundable only against the corporate balance sheet because its cash flow predictability is thin, the leverage capacities of the two are simply not comparable. The position is no different where one project carries a concessional or development-finance tranche while the other is funded entirely on commercial terms. Under such heterogeneity, unlevered return does not measure the relative attractiveness of the two projects at all; it measures the technical efficiency of the underlying assets, and a capital allocation decision is not a technical efficiency decision.
The institutional cost surfaces first in the allocation itself. A project with limited leverage capacity enters the queue because its unlevered return sits a little higher; the project with substantial capacity is held back because it trails by a few points on the same page. Measured at the level of return on equity, the ordering between them can reverse outright, since the debt the second project can support moves both the equity requirement and the after-tax cost of capital onto a different plane. A reversal of this kind is not an isolated error. It is systematic, it repeats in the same direction across successive cycles, and over time it shifts the composition of the portfolio toward assets that are, precisely, harder to finance.
A second cost accumulates in the accounting for tax capacity. Where the interest tax shield is treated as a constant at project level, what escapes notice is that the same shield constitutes a bounded resource at group level: carried-forward losses, accelerated depreciation, incentive-derived credits, and interest deductibility limitations all draw on one taxable base. Where several projects each book the shield to their own account, the aggregate benefit modelled exceeds what the group can actually use. Issuance and arrangement costs shift position in the same picture: structuring fees, commitment charges, guarantee and bonding premiums, external legal and technical advisory expenses, cash trapped in reserve accounts at closing — each of these holds a line in the sources-and-uses table, and most of them hold no line at all in the return calculation that carries the decision.
The same boundary ambiguity produces cost in the opposite direction, and this second form is generally harder to diagnose. A weighted average cost of capital, by construction, already reflects the after-tax cost of debt; where a separate tax shield line is nonetheless added to the cash flow, the identical benefit is counted twice and value drifts systematically upward. A comparable double count arises where the discount rate has been adjusted for financial risk while the free cash flow has not been stripped of interest payments. Both errors originate in the same vacuum: the absence of a written rule fixing which effect lives in the rate and which lives in the cash flow.
A third cost appears at the diligence table. In a sale or investment process, comparing historical projections against realised financing expenses, the counterparty's financial adviser encounters a block of expenditure that never appeared in the model; even where the amount is modest in isolation, the finding is read as a signal about modelling discipline. What that signal typically buys is not a direct price reduction but an additional condition precedent, a broader scope of representations and warranties, a higher escrow proportion, or earn-out thresholds calibrated more conservatively. Put differently, the item excluded from the calculation returns inside the transaction structure.
What neutralises the tendency is institutional architecture rather than individual attention, and it separates into four components. The first is a valuation protocol, fixed on a single page and held constant across every project, specifying which effect is carried in the discount rate and which in the cash flow; the existence of the protocol closes double counting and undercounting at the same time. The second is opening the financing cost record at the mandate letter or first term sheet stage rather than at closing, beginning with estimated amounts and reconciling them against actuals once the facility is drawn. The third is removing tax capacity from the project assumption set and managing it as a bounded resource allocated at group level. The fourth is a review rhythm that requires the ranking to be run once more when financing terms are settled — not to unwind the decision, but to establish whether the decision still yields the same order.
BEIREK's intervention in this area begins with constructing a bridge of named lines between unlevered project value and value accruing to equity, and with keeping that bridge live across the project's life; every line on it — the usable portion of the tax shield, issuance and arrangement expenses, cash immobilised in reserve accounts, hedging costs, security and guarantee premiums — is attached to an owner and a funding source, so that no item falls between the two documents. The accompanying mechanism is the financing cost record opened at term sheet stage and reconciled at intervals after closing between modelled and realised figures; that reconciliation functions not as an audit but as a feedback line calibrating the assumption set for the next transaction. Third, we make it standard for the committee file to show the ranking twice, once unlevered and once with financing effects incorporated: where the two orders agree the discussion is short, and where they diverge the source of the divergence is already legible in the table.
The quality of an investment decision is often revealed less by the accuracy of the cash flow forecasts employed than by whether the boundary of value has been drawn in writing. Financing is not a procurement exercise standing outside the project; it is a constituent part of the project's economics. Where the boundary is drawn so as to exclude that fact, the calculation may be executed flawlessly and still answer the wrong question.