This paper answers a narrow question with wide consequences: does the private equity industry have any institutional mechanism for confirming that the margin improvement built into its underwriting models actually occurred? The decision it should improve is how Investment Committees treat the interval between deal approval and exit — currently a silent period, structurally, in nearly every fund.

The boardroom conversation it should change is the assumption that a rigorously built model is, on its own, evidence that the value it describes exists. The executive habit it should challenge is treating "the model said this would happen" as equivalent to "this happened." It is worth ten minutes of attention because the evidence for this gap no longer comes from outside critique — it comes from the industry's own flagship research.

It should be saved and referenced because every future conversation about exit multiple compression, DPI suppression, or track record durability eventually returns to this same unexamined interval.

Roughly three in four sponsors currently underwrite margin improvement of up to three hundred basis points between purchase and sale. The industry's own benchmark research — Bain and StepStone's 2026 Global Private Equity Report — states plainly that the difficulty is not making this assumption.

It is capturing whether it was ever true, after the deal closes. That sentence, written by the institution whose annual report anchors more Investment Committee memos than any other, deserves a different kind of attention than it has received.

It is not a critique of the industry from outside. It is an admission from within it. The assumption that deserves another look is this: that a margin improvement thesis, once modeled with discipline at entry, requires no further confirmation before it becomes the basis of an exit narrative, a track record, and a buyer's price.

The underwriting model has always been treated as the credible version of the future. Deal teams build it with real analytical seriousness — comparable transactions, third-party operating diligence, management interviews, sensitivity analysis across scenarios. Investment Committees interrogate it before capital commits. This rigor is not performative. It is genuine, and it has historically been sufficient to justify confidence in the number.

Intelligent executives believed the model's discipline at entry was a reasonable proxy for its accuracy at exit, because for most of the last decade, that belief was rarely tested and never expensive when wrong. If the modeled margin improvement of three hundred basis points came in at half that, the gap absorbed itself.

The exit environment did the remaining work. No one had strong reason to ask whether the number in the model and the number in the business were still the same thing.

Two structural shifts have removed the conditions that made this assumption safe to leave unexamined.

First, multiple expansion has stopped being reliable. Bain's own survey found that nearly eighty percent of GPs now expect purchase price multiples to hold roughly at current levels rather than continue expanding. The mechanism that used to absorb the difference between modeled and realized improvement no longer exists in the same form.

Second, and more consequential: the industry has now stated, in its own research, that verifying margin improvement post-close is "the challenge" — not a solved problem, not a minor operational detail, but an open and acknowledged difficulty across a sample of more than a hundred investment professionals. This is not a new discovery about the business. It is a long-standing condition that a changed environment has stopped concealing.

Free cash flow: modeled improvement that was never confirmed may not exist in the cash profile of the business at all, understating true risk relative to what the underwriting implies.

EBITDA quality: reported EBITDA growth and transferable EBITDA growth are being treated as identical in a majority of underwriting models, when the sponsors' own survey data suggests they frequently diverge.

Enterprise value: in three out of four deals, a meaningful share of assumed value rests on an improvement claim with no confirmation event behind it.

Exit multiple: this is where the consequence concentrates. A sophisticated buyer does not discount a company for weak historical numbers. It discounts a company for an improvement story it cannot independently verify — a distinction sellers rarely name, because they rarely recognize it as the reason the multiple compressed.

DPI: exits are delayed, in part, while sponsors informally attempt to firm up a margin story that should have been tested years earlier, rather than confirmed or abandoned on a fixed schedule.

Governance quality: boards currently review the plan at underwriting and the outcome at exit, with almost nothing structured in between.

An experienced Operating Partner sees a distinction that deal teams and boards frequently collapse into one: the difference between a plan that was built well and a plan that was delivered as built.

The board reviews the model. Management manages the business day to day, often without a standing mandate to reconcile actual performance against the specific margin assumptions the deal was priced on. Buyers underwrite the credibility of the improvement, independent of its stated magnitude.

What actually drives transferable value is not the sophistication of the original model — it is whether anyone, at any point during the hold, tested the model against reality before the exit process forced the question.

The differentiated observation is this: rigor at entry has become a substitute for verification during the hold, and the industry has now confirmed, in its own words, that the substitution doesn't hold up.

Sellers present margin improvement as a fact: EBITDA is higher today than it was at entry, and the model anticipated this. Buyers underwrite something narrower and more demanding — whether that improvement can be traced to a specific, confirmable operating event, and whether it will persist under new ownership rather than reverting once the seller's involvement ends.

Confidence increases when a margin claim can be defended by something other than the original model — an operational change, a cost structurally removed, a price increase realized and sustained over multiple periods. Confidence weakens when the only available support is the model itself, restated as history. Most sellers, per the industry's own data, are currently offering the second kind of story without recognizing the difference.

Information sits with portfolio company management, who hold the operational detail behind any margin change.

Authority sits with the deal team, who approved the original underwriting thesis and rarely hold a mandate to revisit it independently.

Incentive sits with everyone involved succeeding together — deal teams by track record, management by continued mandate, Operating Partners by the appearance of a well-run process. No party is structurally incentivized to discover that the assumption failed.

Accountability sits with the fund alone, and arrives four to seven years after the decision that should have triggered scrutiny.

These four dimensions do not converge in a single role. That fragmentation — not a failure of competence anywhere in the chain — is why the gap persists industry-wide rather than being resolved firm by firm.

A three-question diagnostic for any asset at the mid-hold mark or approaching exit:

An asset that cannot answer "yes" to at least two of the three should be treated, internally, as carrying unverified enterprise value — regardless of how confidently the improvement appears in the current model.

Sponsors should treat margin verification as a governance requirement at the mid-hold mark, not an artifact discovered during exit preparation.

Operating Partners should hold a standing mandate to test underwriting assumptions independently of the deal team that made them.

CFOs should distinguish, in internal reporting, between margin improvement that has been traced to a specific event and margin improvement that remains an inference from trend data.

Boards should ask, at any point past year three of a hold, whether the original margin thesis has been tested against reality — not merely restated in updated projections.

Investment Committees should require the Underwriting-Realization Test, or an equivalent standard, before treating any margin assumption as a settled input to exit valuation.

None of this is a tactical implementation question. It is a governance question: whether the organization has decided that unverified assumptions are acceptable inputs to a track record.

"A model that has never been tested against reality is not evidence. It's a forecast wearing a track record's clothes."

The page that belongs in the next board packet is the Underwriting-Realization Test, applied to the three most recent exits or the three longest-held current assets.

The question that should appear in the next operating review: for each asset past year three, can management trace the realized margin improvement to a specific event, or only to the difference between two EBITDA figures?

The assumption that should be re-underwritten: any margin thesis currently supported solely by the original entry model.

This paper adds a citable, primary-source anchor to two existing frameworks — the Two Scorecards Problem and the Transferability Test — neither of which previously had a direct industry admission behind them.

It makes permanently referenceable the distinction between modeled and verified EBITDA, and it establishes the Underwriting-Realization Test as the pillar's first concrete diagnostic tool.

Future research on exit multiple compression, DPI suppression, and track record durability can now build directly on this paper rather than re-arguing the underlying gap.

Before reading, the executive likely believed: a well-built underwriting model is sufficient justification for the margin assumptions carried into an exit narrative.

After reading, the executive should believe: an underwriting model, however rigorous, is a forecast until independently tested — and the industry's own research confirms that testing rarely happens.

Before reading, the executive likely governed by: reviewing the model at entry and the outcome at exit, treating the interval between as management's domain.

After reading, the executive should govern by: requiring independent verification of margin assumptions at a fixed point during the hold, owned by a role distinct from the one that built the original model.

The video persuades the audience to feel the discomfort of the admission — that the industry confirmed this gap about itself, and that the audience's own deals are implicated. This paper explains, with more precision than spoken delivery allows, exactly where the economic consequence concentrates (exit multiple), why the four-part ownership gap persists structurally rather than by individual failure, and what a board or Investment Committee should concretely require going forward. Only the written format can carry the Underwriting-Realization Test as a durable, reusable diagnostic — future publications should cite this paper specifically when invoking that framework, rather than the episode.

A Managing Partner discovering this paper in 2033 would find the core argument unchanged: the observation does not depend on the 2026 multiple environment, the current fundraising climate, or any temporary market condition — those are cited only to explain why the gap became newly visible, not as the substance of the claim.

The enduring principle — that underwriting discipline at entry does not substitute for verification during the hold — has no expiration date, and the Underwriting-Realization Test remains applicable to any hold period, in any market environment, indefinitely.