Compare two files that reach an investment committee eighteen months apart in the same geography, the same technology and with a broadly similar counterparty profile, and the second will typically clear at a visibly lower return hurdle than the first, even though nothing has changed in the technical risk of the asset, the complexity of the construction programme, or the degree of offtake concentration on the revenue side. What has changed is the number of transactions that closed in the intervening eighteen months and the levels at which those transactions cleared. No member of the committee proposes, in that session, that the price of risk be reduced; the proposal is always framed more narrowly than that — the market is clearing here, and this asset should not be lost. Each individual step looks rational precisely because it represents only a marginal departure from the step preceding it.
On the credit side the same pattern advances even more quietly. Where the spread has landed in basis points is minuted and argued over; which covenant headings were dropped, how a reserve account moved from twelve months of debt service to three, and which assets the security package now leaves outside its perimeter are rarely debated with equivalent granularity. Compensation for risk is never price alone — it is the resultant of price, protection and information rights taken together, and the second and third of those components can erode while the first holds perfectly still. An institution may therefore have materially reduced the compensation it receives for risk without ever having moved its stated return target.
The mechanism underlying this erosion is **risk-premium compression** — the narrowing, during optimistic cycles, of the return spread demanded for bearing risk to a level that cannot be sustained across the asset's life — and it operates less as a forecasting error than as a calibration drift. The decision-maker does not measure the absolute level of risk on an independent footing, and could not do so in any case, since no absolute footing exists. What is used instead is the nearest observable reference: comparable transactions that closed in recent quarters. Once each new transaction anchors to the previous one, the reference series becomes a self-feeding chain, and the original justification at the head of that chain turns invisible within a handful of links.
There is a functional side to this mechanism, and ignoring it produces a mistaken diagnosis. Leaning on market reference is a shortcut that compresses decision time in an environment where information is expensive and that blunts idiosyncratic bias; an institution attempting to price every file from first principles, on its own independent risk footing, would likely close nothing on schedule and would forfeit comparability across its own portfolio. The problem lies not in the shortcut but in its validity condition: a reference series carries information only while the conditions generating that series hold constant. Where abundant liquidity, deployment pressure created by a fundraising cycle, or policy incentives directed at a particular asset class are present, the series no longer measures the price of risk but the crowding of capital, and conflating the two is the principal source of compression.
A second layer originates in the institution's own incentive architecture. Investment team performance becomes visible through capital deployed and transactions closed, while the opportunity cost carried by undeployed capital is not measured with comparable visibility. The cost of passing on a transaction surfaces immediately, concretely and in a form that can be attributed to a named individual; the cost of entering at too thin a margin accrues over the life of the asset, on a horizon that frequently exceeds the tenure of the team that made the decision. That asymmetry pushes decision-makers predictably toward thinner margins, and the movement reflects not individual weakness but the direct consequence of how performance is measured.
The institutional cost accumulates, at the first stage, not in the income statement but in the flexibility line. An asset structured on a thin margin can deliver its targeted return in the base case; the same asset, however, carries no buffer capable of absorbing a single adverse deviation — a commissioning schedule slipping by one quarter, operating expenditure running above inflation, an offtaker's credit profile stepping down one notch. The genuine loss embedded in such a structure is not a lower expected return but the absence of manoeuvring room in the lower tail of the distribution around that return. Where a reserve account holds three months rather than twelve, the institution finds itself, in a poor quarter, caught between calling fresh equity from the sponsor and defaulting on debt service, neither of which was priced at the time of the transaction.
The second cost appears at portfolio level and with a lag. Where an institution concentrates the capital deployed during a compression window into a single vintage, a persistent gap opens between the portfolio's blended return and that vintage's return, and subsequent transactions entered at wider margins close that gap only partially. At fund level this means that a meaningful share of committed capital was bound at the narrowest point in the cycle; on a corporate balance sheet the same phenomenon surfaces as tension between the carrying value of investment assets and the discount rate applied in revaluation work, typically emerging in a footnote to an independent valuation report.
The third cost sits in the structure of the counterparty relationship. In a transaction structured on a thin margin, the institution's contractual leverage has thinned in parallel; to the extent that information rights, consent thresholds and step-in mechanisms have been narrowed, the institution can only come to the table after an operational deviation has already occurred, and from a weakened position. What determines bargaining power in a renegotiation is not the price paid at signing but the rights retained at signing, and those rights are precisely the items most readily surrendered during optimistic cycles under the rationale of remaining competitive.
The mechanism that neutralizes this tendency is decision architecture rather than individual prudence, and it has four separable components. The first is an annual rejustification of the discount rate and the target risk compensation on a cycle-independent footing — the technical life of the asset, the credit profile of the counterparty, the regulatory stability of the jurisdiction and the historical range of deviation — with that justification held in writing. The second is that every file records, in the minutes, how far the transaction price departs from that independent footing and on what grounds; departure itself is not prohibited, leaving departure unrecorded is. The third is that the non-price components of risk compensation — covenant headings, reserve months, collateral scope, information rights — are tracked on a separate checklist and rendered comparable across files. The fourth is the institutionalization of a counter-argument role: a person independent of the deal team, tasked solely with defending the lower tail of the structure.
BEIREK's intervention on capital-intensive projects is constructed precisely on this surface. Rather than reducing risk compensation to a single return figure, we maintain, across the projects we manage, a risk-compensation register that records the price, protection and information-rights components separately; every structural concession is entered into that register together with the competitive condition under which it was granted and the approval by which it was authorized, and the register is placed before the investment committee as a single document ahead of FID. What the committee then sees is the sum of ten concessions each of which appeared reasonable in isolation — which is the only moment at which compression becomes institutionally visible.
Alongside this, on projects where we manage the development and financing workstreams, we rebuild the rationale for the discount rate on a calendar rhythm decoupled from the cycle, and pair that rhythm with a stakeholder pre-mortem: which single deviation in the lower tail exhausts the reserve account, which covenant heading trips first, and at what point the sponsor faces a call for fresh equity are established in writing before closing. This exercise does not make the structure conservative by default; it separates entering a thin margin knowingly and with that thinness priced from entering it out of habit.
The most direct way to gauge an institution's risk culture is to look not at how the discount rate has moved across files approved over the past three years, but at where the justification for that movement is written down. Where the justification rests solely on market reference, the institution is not setting the price of risk but carrying it; and a carried price remains on the institution's own balance sheet once the cycle turns.