The Real Cost of a Ten-Day Close: What Time Pressure Does to the Baseline

It is a Thursday afternoon and the broker has called for best and final by Monday. The seller wants exchange within the week that follows and completion as soon as the fund can draw. The fund’s limited partnership agreement gives investors ten business days’ notice of a capital call, which is the common period in venture and private equity documents, and the completion date is worked back from that notice. The team has, in practice, a few days to fix a price and ten more to be sure of it, and any check that does not fit inside those days will be done after the money has moved or not at all.

Ten days is not a figure chosen for effect. It is the notice period written into most LPAs, and funds-of-funds are advised to hold their own period to seven days or fewer so that they can meet a call from an underlying manager.

One clarification before going further. The phrase “ten-day close” also belongs to the accounting calendar, where APQC’s cross-industry benchmark puts a bottom-quartile monthly close at ten or more calendar days against 4.8 for the top quartile. That is a different ten days, about book-keeping cadence, and it is not the subject here. This paper is about the acquisition close, the point at which a set of underwriting inputs becomes a purchase price, and about what happens to those inputs afterwards.

The claim is that the visible cost of compression, the overtime and the lost weekend, is the smaller part. The larger part is structural: the window decides which figures are verified before they become the price, and the unverified figures become the baseline against which the asset is budgeted, valued and reported for the rest of the hold. The Smart Bricks Research Lab has been working through why that baseline is so rarely re-examined, drawing on the decision-research and appraisal literatures and on the regulator and insurer records of where the error finally surfaces.

The intuitive model of a compressed close is that every check is done faster and a little worse. Decision research says something more specific. In a series of experiments on choice among alternatives with many attributes, Payne, Bettman and Johnson found that people under time constraint do not run their usual procedure at higher speed; they switch strategy, moving to heuristics that examine some attributes and ignore others. Several of those heuristics performed better than a full comparison that had simply been cut short, which is why people adopt them. The adaptation is sensible. Its cost is that the attributes dropped are the ones that take longest to reach.

In an acquisition the slow attributes are the document-heavy ones. The rent roll is a table; the executed leases and their amendments are a stack of files in a data room whose folder structure was designed by the seller. The technical report has a summary; the consultant findings behind it are appendices. The term sheet is two pages; the credit-agreement draft is two hundred. Under compression the team reads the table, the summary and the two pages, and a specific set of substitutions follows, of which two illustrate the shape.

The seller’s rent roll stands in for the leases, so passing rent is taken from a document prepared by a party with an interest in how it reads, and where a lease has been amended the base lease is abstracted and the amendment missed, so escalation steps and break options are read from the instrument that no longer governs. The survey’s cost opinion is read as a capex figure, although the ASTM standard under which such surveys are written describes its cost estimates as order-of-magnitude budgetary projections, and the contingency that phrase implies never reaches the model. Debt, market inputs and the model itself follow the same pattern, and Exhibit 1 sets out where each substitute lands.

Nothing external forces the dropped check back into the window. So the triage that decides which checks survive is made inside the team, late in the evening, without a rule. It is also made without a record: the file will show that a rent roll was reviewed, not that the leases were not.

That is the asymmetry worth being honest about. Analyst overtime is visible, budgeted and recoverable. The triage decision is invisible, unbudgeted and, as the next sections argue, the only cost of the compressed close that compounds.

• 10 business days

  LPA notice period before called capital is due, for many private equity and venture funds; a bridge facility draws the same day

  Morgan Lewis, Capital Calls (2015)

• 2 months

  Average time to write and review an investment memo, ranging from two weeks to a year, across 54 institutions

  Addepar and Stanford Long-Term Investing (2023)

• 5 working days

  Window after completion within which property and tenant financial information should reach the buyer

  RICS, Service charges in commercial property, Appendix D

• 5 and 10 years

  Maximum tenure of the responsible valuer and of the firm on UK regulated purpose valuations, from 1 May 2024

  RICS Red Book UK national supplement

• 1% of NAV

  Tolerance below which a Luxembourg vehicle investing in buildings need not correct or compensate a NAV error

  CSSF Circular 24/856

None of these clocks was set to give the buyer time to verify the numbers. The verification either fits inside them or does not happen.

Source: Morgan Lewis (2015); Addepar Research Brief with Stanford LTI (2023); RICS Service charges in commercial property (2018, reissued 2023); RICS Red Book UK national supplement (2023); CSSF Circular 24/856 (in force 1 January 2025)

Follow one input after the close. The passing rent accepted from the seller’s rent roll is priced into the bid, and on completion it leaves the deal team and begins a second life. It passes through five processes before the asset is sold, from the business plan presented to the committee through the year-one budget, the first external valuation and the fund’s NAV to the hold-or-sell decision, and at every one of them it is consumed as a fact rather than re-derived from the lease it came from.

Each of those processes has a framework governing the cadence and format of its output. INREV expects an annual asset-level plan, the NCREIF PREA Reporting Standards expect quarterly asset reporting, the RICS Red Book governs the valuation, and Luxembourg and UK regulators govern the NAV. None requires the input to be traced back to the executed document or reconciled to the acquisition case. The business plan is drafted to win the committee’s approval and then filed apart from the reporting that could later contradict it; the handover from deal team to asset manager typically re-keys the data room by hand rather than carrying the model’s inputs across with their sources.

The gearing between a single line and the value it supports is easy to state. Take an asset whose net operating income has been overstated by USD 75,000, because one tenant’s amendment reduced the rent and the rent roll was not updated. Capitalised at 6%, that is USD 1.25 million of indicated value. More generally, a 5% error in NOI moves an income-capitalised value by roughly 5%, and it moves it in every valuation that starts from the same rent roll. This is arithmetic rather than a statistic, and it is ours, but it is the reason a rent-roll line deserves the same attention as the yield.

Capital expenditure has the same gearing in the other direction. When a survey’s cost opinion is read as a firm number and the contingency it implies is never carried, the maintenance the asset needs does not reach the reserves line, and operating income is overstated by the difference. The pattern has been measured outside real estate: across listed companies, Peddireddy estimates that the median firm recognises depreciation and amortisation about 25% below its estimated maintenance capital expenditure, overstating operating income by around 7%. That is a cross-sector finding about listed equities, not a measurement of property, and it is cited for the shape rather than the number. The analogue in a deal is the survey finding that reaches the file but never the model.

Input	Set at close from	Becomes the baseline for	How it compounds	What can correct it
Passing rent and rent roll	Set at close fromSeller's tenancy schedule, partly reconciled to leases	Becomes the baseline forLedger income, budget, valuer's passing rent	How it compoundsCapitalised into value; indexed as the base of every escalation	What can correct itLease-by-lease reconciliation; a tenant dispute; the next buyer's diligence
Operating costs and recoveries	Set at close fromSeller's operating statement	Becomes the baseline forFirst-year budget, service charge apportionment	How it compoundsRecoveries mis-billed year after year; NOI overstated at every valuation	What can correct itService charge reconciliation within four months; tenant audit
Capex allowance	Set at close fromConsultant's order-of-magnitude cost opinion	Becomes the baseline forBusiness plan capex programme, reserve line	How it compoundsDeferred works compound; residual value at sale absorbs the gap	What can correct itCondition survey at refinance or sale
Market rent and growth	Set at close fromBroker guidance, seller's projections	Becomes the baseline forReversion in model and valuation, leasing targets	How it compoundsAnchors the valuer; drives ERV in every subsequent draft	What can correct itActual lettings; valuer rotation after five years
Exit yield	Set at close fromComparable evidence assembled under deadline	Becomes the baseline forTerminal value, hold/sell analysis, IRR reported to LPs	How it compoundsApplied to compounded NOI at sale; dominates the residual	What can correct itThe sale itself, once in five to ten years

Each input leaves the deal team as an assumption and enters the operating stack as a reference value. Nothing downstream is designed to ask whether it was ever verified.

Source: Smart Bricks Research Lab

The natural objection is that markets correct errors. A price set on an overstated rent roll should be revealed by the first valuation, the first budget variance or the first investor question. The evidence on how valuations are actually formed says otherwise, and it comes from three directions.

The first is anchoring on a supplied figure. Northcraft and Neale gave practising real estate agents a ten-page information pack on a house that had been independently appraised at USD 74,900, and varied one line: the listing price. Agents shown a listing of USD 65,900 appraised the property at USD 67,811 on average; agents shown USD 83,900 appraised it at USD 75,190. One changed line moved the average professional opinion by USD 7,379. Students moved further, from USD 63,571 to USD 72,196 across four anchors, but the experts moved too, and very few of them named the listing price as something they had considered. The study is residential and experimental, and it should be read as such. What it shows is that a competent professional with the full pack in front of them still starts from the number they were given.

The second is anchoring on one’s own prior figure. Clayton, Geltner and Hamilton examined individual commercial appraisals and found that appraisers valuing the same property in consecutive periods weight their previous appraised value, producing more lag than appraisals of properties they were seeing for the first time. Their recommendation was to rotate appraisers. This is the mechanism by which an acquisition price becomes a valuation and a valuation becomes the next one: the valuer is adjusting a figure that already exists rather than re-deriving the asset’s worth from its documents each period. The RICS Red Book’s UK national supplement, effective from May 2024, now requires that the responsible valuer on a regulated purpose valuation rotate after at most five years and the firm after ten, with a three-year break. That is the profession’s institutional remedy for the anchoring its own literature identified, and it runs on a cycle as long as many funds’ holding periods.

The first valuation after completion starts from the price paid, and every valuation after that starts from the last one.

Smart Bricks Research Lab

The third is anchoring on cost. Bokhari and Geltner, working from commercial transaction data, found that loss aversion plays a major role in commercial property pricing, that asking prices appear to anchor buyers’ valuations, and that larger and more experienced institutions exhibit at least as much of the behaviour as smaller ones. An owner anchored on cost does not re-underwrite cost. The business plan that would reveal the acquisition case to have been optimistic is the document the owner wrote, and the variance against it is explained one quarter at a time.

How large is the gap that eventually closes? MSCI’s comparison of valuations with subsequent sale prices for 2025 puts the weighted average absolute difference at 7.6% globally and 7.0% in the United Kingdom, on 486 UK sales, against a ten-year global average of 9.5%. The weighted average signed difference was 0.1%, which means the valuations carried no aggregate bias; valuers were not systematically high or low. That matters for the argument here. The correction is an asset-level event, of the order of seven percent in either direction, and the market applies it at the point of sale. Until then the asset is carried at a figure that descends, through a chain of anchored valuations, from the price a deal team fixed in a compressed week.

If the hold-period machinery does not re-test the acquisition case, something else eventually does, and it is worth being precise about what.

The first is a warranty claim. Marsh’s 2024 report on transactional-risk insurance claims found that breaches of financial-statements warranties accounted for 45% of the payments insurers made in 2023, the largest single category by value. Cumulative payments to Marsh clients since 2018 passed USD 1 billion net of retention. Tax breaches, by contrast, were about 20% of notifications and 2% of payments. The data covers Marsh-placed policies across all M&A rather than real estate alone, but the direction is instructive: the misstated number at signing, rather than the structural or legal defect, is what insurers pay for most.

The second is a NAV error. Under Circular CSSF 24/856, in force from January 2025, a Luxembourg investment vehicle investing primarily in assets other than equities and bonds, the category that includes real estate, has a NAV calculation error tolerance of 1.00%. Above that, the error is compensable to the investors affected and notifiable to the regulator. Vehicles reserved for professional investors may document a higher threshold, up to 5%, but only on the basis of a written analysis and never by default. A rent-roll error that moves one asset’s value by the 5% of the illustration above, in a fund holding a handful of such assets, is within reach of that threshold.

The third is a dealing suspension. The UK Financial Conduct Authority’s rules for non-UCITS retail schemes investing in inherently illiquid assets, in force since September 2020, require a fund to suspend dealing where its standing independent valuer has applied material uncertainty to immovables making up more than 20% of scheme property, unless the manager and depositary decide otherwise within two business days. Valuation confidence is written into the conditions under which investors can redeem at all.

The fourth is time itself. In February 2016 the Ontario Securities Commission approved a settlement with CI Investments over interest that had been accrued but not recorded in the NAV of several of its funds. The amount was CAD 156.1 million. The error persisted from December 2009 to June 2015, five and a half years, before the firm found it, reported it and returned the money to around 360,000 investors. These were mutual funds rather than property, and the lesson concerns the interval rather than the asset class: an unrecorded item sat inside a reported NAV through more than twenty quarterly cycles without any of them catching it.

Whether systematically optimistic acquisition cases show up in aggregate returns is harder to say, and the one study that comes close should be read with care. Pagliari compared value-added and opportunistic fund returns with a core index levered to the same volatility over 1996 to 2012, and found that value-added funds underperformed by about 180 basis points a year while opportunistic funds roughly matched it. That is a measurement of strategy returns net of fees, and it says nothing about diligence depth. But a value-creation case that is underwritten at acquisition and, on average, not delivered is what a systematically optimistic baseline would look like from the outside.

• 29%

  Share of USD 650 billion of CMBS loans (2013 to 2019) whose underwritten income exceeded actual property income by 5% or more

  Griffin and Priest, Journal of Finance (2023)

• 25% and 7%

  Depreciation recognised by the median US firm below its estimated maintenance capex, and the resulting overstatement of operating income; cross-sector finding

  Peddireddy, Columbia CEASA (2021)

• 8 months

  Interval after which expert appraisers updating their own valuation made insufficient adjustment from their previous figure

  Diaz and Wolverton, Real Estate Economics (1998)

• 45%

  Share of 2023 transactional-risk insurance payments arising from breaches of financial statements warranties

  Marsh, Global Transactional Risk Insurance Claims Report 2024

• 72%

  Share of 2005 to 2007 vintage US closed-end real estate funds that underperformed the index on both levered and unlevered bases

  MSCI (2022)

Where anyone has measured what happens to income and cost inputs at origination, the error runs one way, and the review that follows under-adjusts rather than corrects.

Source: Griffin and Priest (2023); Peddireddy (2021); Diaz and Wolverton (1998); Marsh (2024); MSCI, Robson and Harmse (2022)

None of this argues for a longer close. The market sets the window, and a firm that will not bid inside it does not buy. What a firm can do is decide, before the process begins and at the level of the mandate rather than the deal, which figures a close may not proceed without. Call it the no-go set. It is short, it is written down, and it exists so that the triage the team makes on Thursday night is a rule rather than a judgment.

The candidates for the set are the inputs with the greatest gearing into the hold. Passing rent, reconciled line by line to the executed leases and their amendments, with break options and rent reviews priced rather than noted. Capital expenditure, with every consultant finding in one schedule and the survey’s cost opinion carried as the order-of-magnitude estimate it is, with a stated contingency. Debt, with covenant definitions and cash-sweep triggers taken from the credit-agreement draft rather than the term sheet. And the NOI definition itself, stated in writing, so that the year-one P&L can be compared with the acquisition case on the same basis. If a process does not allow those four to be completed, the firm has decided in advance that it does not bid, or that it bids at a price that reflects what it could not check. Everything else may be triaged.

The second decision is to save the acquisition case as a baseline rather than let it dissolve into the business plan. The model that priced the deal is versioned at close, each material assumption carries its as-at date and a named owner, and the business plan and year-one budget are derived from that file rather than re-typed from it. This is the join that is usually missing at handover. Without it, drift from the acquisition case is remembered by whoever was in the room; with it, drift is measurable.

The third is a re-underwrite at month twelve. The ten most material acquisition inputs are compared with outturn, the variance is attributed to a cause and an owner, and the count of inputs that moved beyond a written tolerance is reported to the committee. No framework will require this, since none compares the plan with the case it came from. The count is therefore an instrument a firm has to build for itself, and it is the instrument that turns the compressed close from an unrecorded bet into a measured one. The first year’s count will be uncomfortable. That is the finding.

What this does to the ten-day close is worth stating plainly. It does not lengthen it. It converts a triage made under pressure into a rule made in advance, and it moves the point at which the dropped checks are paid for from the sale, years later and through the next buyer’s diligence, to month twelve, inside the firm and in its own reporting.

1. 1

   Day 0, completion

   Underwriting model writes to the ledger; rent roll, costs, capex and yields become reference values • No framework requires a verification state to travel with the value

2. 2

   Day 5, handover

   RICS expects property and tenant financial information to reach the buyer within five working days • The clock exists for service charges, not for re-verifying the underwriting

3. 3

   Month 3, first valuation

   Valuer instructed on the acquisition rent roll and price • Diaz and Wolverton: insufficient adjustment from the prior figure; the price anchors the value

4. 4

   Year 1, business plan review

   Actuals compared to pro forma • Variance is attributed to the market, not to the baseline it is measured against

5. 5

   Years 1 to 3, warranty window

   W&I claim possible where a warranted figure was wrong • Pays where a buyer finds the error and has a warranty; 45% of payments are financial statements breaches

6. 6

   Every quarter, NAV

   Fund accounting consolidates valuation into NAV • Below 1% of NAV no correction is required; the compounded error is the NAV

7. 7

   Year 5, valuer rotation

   Responsible valuer must change on UK regulated purpose valuations • The first independent look at the baseline, half a hold period after it was set

8. 8

   Year 5 to 10, sale

   The next buyer's diligence tests the rent roll against the leases • The only mechanism that reliably fires, and it prices the error into the residual rather than correcting it

Every point at which a wrong acquisition input could be caught, and why most of them do not catch it. Read down the right-hand column: the corrections are rare, thresholded or arrive after the residual has been set.

Source: Smart Bricks Research Lab, drawing on RICS Service charges in commercial property Appendix D; Diaz and Wolverton (1998); Marsh (2024); CSSF Circular 24/856; RICS Red Book UK national supplement (2023)

The trade-off is genuine. Brokers route early looks to bidders who respond quickly, and a firm known to be slow sees fewer compressed processes and wins fewer of the ones it sees. A no-go set is a set of terms on which the firm will not proceed, and there will be deals it loses to a bidder without one. The argument here is for a defined set of terms inside whatever window the market allows, so that the deals a firm wins on ten days are the deals whose baseline it can defend, and it is an argument against slower closes as much as against unconditional fast ones.

The cost has a shape, and most of it is a redirection of time already spent. The no-go set is a mandate document written once and reviewed annually. The versioned baseline is analyst time at close that is currently spent re-keying the data room into asset management systems, redirected into carrying the model’s inputs across with their sources. The month-twelve re-underwrite is roughly a week of one analyst per asset per year, and it should be budgeted as a named role, because a re-underwrite nobody owns is the first thing dropped when the next acquisition compresses.

The evidence has limits, and they should be stated. MSCI’s 0.1% weighted average difference means valuations carry no aggregate bias, so this is an asset-level argument and not a claim that the market misprices. Northcraft and Neale studied houses in an experiment; Clayton, Geltner and Hamilton studied the smoothing of appraisal indices; Peddireddy studied listed companies across sectors; Pagliari measured strategy alpha net of fees and said nothing about the diligence behind it. The Marsh, CSSF and Ontario figures come from M&A insurance, fund regulation and mutual funds respectively. Each shows one link of the chain. No published study measures the link the argument rests on, the relationship between verification depth at acquisition and outturn over the hold, and this paper does not claim one. The month-twelve count is proposed partly for that reason: a firm that keeps it for a few years will hold the first dataset on the question.

The window, then, is not the problem. Ten days is the market’s number, and the fund’s own documents confirm it. The problem is the triage inside the window, made without a rule and without a record, whose result is written into the baseline the asset will be measured against until it is sold. A firm cannot lengthen the clock. It can decide in daylight what the clock is not allowed to cut.