Worked structure four · Structural Risk Review
An exit narrower than the term.
The asset is illiquid, the fund has a term, and the number of parties who could ever buy the asset is small and knowable at entry. Every structural decision taken at acquisition decides something about the exit, and almost all of them are taken by people thinking about the acquisition.
The structure is designed backwards from the exit the asset can reach, not forwards from the entry.
Worked structures · fifteen rooms
- 01An existing platform
- 02An ineligible asset
- 03An investor class
- 04A narrow exit
- 05A repeatable structure
- 06A co-investment
- 07A continuation
- 08Unavailable security
- 09The seed terms
- 10A change of control
- 11Drawn elsewhere
- 12An in-kind distribution
- 13A change of domicile
- 14An investor's perimeter
- 15A strategy, no vehicle
01 · The transaction as it arrives
The asset was bought by one set of people. It is sold by another.
Four arrivals. One is cheap, one is a repair carried out while there is still time to carry it out, one is a negotiation with a party who already knows the date the fund has to be out, and the fourth is the disguise the other three travel in.
The rarest of the four and the only one where every route is still open. The exit is a paragraph in a design document rather than a transaction, and it costs a paragraph.
Whether an intermediate entity is still fit to be sold as shares. How the asset is valued. Each is small on its face. What each one tests is whether the second exit route is still alive, and the answer is frequently that it stopped being alive in a year nobody now remembers.
The fund is inside its final extension, the holders have been given a date, and the asset has one plausible buyer. What is wanted here is not design. It is a comparison of two unattractive routes, drawn to the same depth, with a full account of what each one costs the register.
Almost never "design our exit". It arrives as a valuation policy to be reviewed, a lender consent to be obtained, or one line in a diligence list nobody can close. The structural question sits behind whichever of those came through the door, and it is the same question in all four.
The asset in this class is not exotic. It is the ordinary illiquid holding of a private-market fund. What makes it this case is that the set of parties able to take it off the fund can be written on one line, and that line was already short on the day the fund bought it.
No manager, asset, counterparty, value or date appears in this room. What transfers from one transaction to the next is the shape of the problem, and the shape is what is set down here.
02 · The structural problem
Designing an exit for an asset whose buyer universe is smaller than the fund's term.
Two counts decide the exit and nobody puts them on the same page. One belongs to the asset and one to the constitution. They are held by different people, and they are compared for the first time by the party sitting across the table at the end.
An exit is not an event at the end of a hold. It is a property of the structure built at the beginning, and by the time anybody is thinking about it most of its terms have already been agreed with people who are no longer in the conversation.
The specific difficulty is a mismatch of two counts that nobody puts on the same page. On one side, the number of parties who could lawfully and practically buy this asset, or the entity that owns it, at any price. On the other, the number of months left in the fund's term, including whatever extensions the constitution permits and whose consent each extension needs. Where the first count is small, the second count stops being a schedule and becomes a term of the sale, and the buyer knows it.
Why the two counts stay apart
The first is an asset fact, produced by the asset's own jurisdiction, by the licences attaching to it and by the ownership rules on its register. The deal team establishes it at acquisition because the purchase cannot be made without it, and it is then filed with the acquisition papers and not read again.
The second is a fund fact. It sits in the constitution, it is administered by people who were not in the acquisition, and it is reported to the register as a date. Nobody produces it as a number of months against a named asset, because the term belongs to the fund rather than to any one thing the fund holds. The comparison takes an afternoon, and no single person on the transaction owns both numbers.
The party across the table does it anyway. A buyer approaching an asset with a small universe reads the fund's own reporting and works out the date at which the seller stops being a seller by choice. The manager arrives at that meeting holding one of the two numbers. The buyer arrives holding both.
The problem is not that the asset is illiquid. Illiquidity was priced at entry and the holders agreed to it. The problem is that the term was priced as a schedule and is enforced as a term of the sale.
03 · What binds
Four of the six were settled by the manager, and one is not settled by anybody.
Six. Four of them are settled by decisions the manager itself takes at or before acquisition, one was settled at fund formation, and one is not settled by anybody.
The term and its extensions, written in the constitution, with the consent each extension requires. The transfer provisions attaching to the asset or to the entity that owns it: pre-emption rights, consent requirements, and change-of-control provisions in the asset's own contracts. The security package, under which a change of control is an event the lender answers on its own timetable. The local rules on who may own the asset, which decide the buyer universe before any commercial question is asked. The valuation question, which decides whether a price can be produced by a process or only by an opinion. And what holders were told about liquidity at subscription, which is the one constraint with no document to enforce it and the largest consequence when it is missed.
Each one, by source and by what can move it
The constitution, fixed at formation. Movable by a vote, which means movable at a price, and the price is paid in holder confidence rather than in anything the model carries.
Contracts signed at acquisition, most of them drafted by somebody else and accepted because agreement was cheap that day. Each right installs a party with a say in the exit, and that party has no reason to be helpful at the moment it is asked.
The facility. Security over the chain fixes who has to be paid, released and satisfied before title moves, and it answers on the lender's timetable rather than the transaction's.
The law of the place the asset sits and the requirements of the register that records title there. The only constraint on the list that no party to the transaction can move at all.
The fund's valuation policy and the appointments made under it. Whether a price comes from a process or from an opinion is settled by who was appointed, and when.
Nothing enforceable. No document carries it and no adviser holds it. It is enforced by the register's willingness to commit to the next vehicle.
Five of the six can be produced from documents by somebody who knows which documents to ask for. The sixth is produced from what people remember being told.
04 · The architectures considered
Four routes exist at the start, and the entry decides how many are left.
Four exit routes, and the structure at entry decides which of them are still available.
A sale of the asset itself, into whatever universe the asset's own jurisdiction permits. A sale of the shares of the entity that owns it, which reaches a wider universe and requires that the entity was built to be sold from the day it was formed. A continuation, in which a new vehicle buys the asset and the existing holders choose between cash and a rolled interest. A partial realisation supported by financing, which is not an exit and is often described as one.
The second route is the one that most often disappears without anybody noticing, because it depends on an entity having been clean, resident where it said it was, and free of contracts that a buyer of shares would inherit. Those are conditions maintained over years, not conditions arranged at the end.
What each route reaches, and what closes it
Reaches only the parties the asset's own register and licensing rules accept as an owner, and nothing done inside the structure widens that set. Closed by nothing, which is why it survives neglect, and felt at full strength when the universe is small.
Reaches every party willing to own shares in a company, which is a materially larger set, and carries the contracts, the licences and the history with them. Closed by a residence that lapsed, a second asset acquired, an audit not taken, or one clause signed in the entity's name.
Reaches capital that wants the asset without wanting the fund, and gives the holders a choice rather than a date. Closed by a price that cannot be defended to the people leaving and the people staying at once, because the same number serves both.
Returns capital without moving the asset, and the term goes on running underneath it. Where it buys time, the time has to be spent on one of the first three. Where it replaces them, it adds a lender to the consents the eventual exit needs.
The same problem drawn as a sequence. Three of the five steps are taken at entry, and none of them can be taken again.
- 01 The holding form chosen Whether the exit will sell an asset or sell the entity that owns it is settled here, years before anybody sells anything, by people whose whole attention is on the purchase. One-way
- 02 The consents accepted at entry Pre-emption rights, transfer consents and change-of-control provisions taken on at acquisition. Each is a party who must agree to the exit, and each was granted at a moment when agreement was cheap to give. One-way
- 03 The security granted Security taken over the chain fixes who has to be paid, released and satisfied before title moves, and it fixes it in a document the exit cannot amend on its own timetable. One-way
- 04 The hold The years in which the second exit route is either maintained or quietly allowed to lapse. The only phase on this bar where a decision taken at entry can still be repaired, and the only one with no event in it to make anybody look.
- 05 The exit attempted The buyer universe is now whatever the first three steps left, the timetable is whatever the term left, and both facts are visible to the party across the table.
Three of the five are taken at entry and cannot be taken again on different facts. The fourth is the only place where a mistake in the first three can still be repaired, and it is the phase with no event in it to prompt anybody to look. That is the argument for drawing the exit before the acquisition rather than after it.
05 · The critical dependencies
A condition held for a decade, not an item done in the final quarter.
Three, and each of them is a condition held over the whole hold rather than an item done at the end.
That the entity intended to be sold remains saleable: its residence maintained where the design assumed, its accounts capable of surviving diligence, and its contracts free of terms a share buyer would refuse to inherit. Owned by the manager, tested annually, cheap to maintain and expensive to restore.
That every consent the exit needs has been identified and its holder named. Owned by counsel, read from the asset contracts and the facility rather than from memory, and mapped at entry rather than at sale.
That a price can be produced by a process rather than by an opinion. Owned by the manager through the appointment of an independent valuer and a process party independent of it. A price a process produced can be shown to the people who will test it. A price an opinion produced is the manager's own mark with a covering letter on it.
What each costs to hold, and what it costs to restore
The first costs an afternoon a year. The entity is put against four questions: is it still resident where the design assumed, does it hold the asset and nothing else, are its accounts on a basis a buyer will accept, and has anything been signed in its name that a share buyer would refuse to inherit.
Restoring it later is different work, because the years already in the entity cannot be re-lived. The most that can be done late is to price the defect and disclose it, and the buyer prices it too, from the other side, with the term visible.
The second costs a list, drawn at entry from documents already on the file rather than discovered at exit under a timetable. The third costs an appointment made while nobody needs it, because a valuer appointed in the year of a continuation is a valuer whose independence is a question.
A dependency with no owner is not a dependency. It is a hope with a deadline on it.
06 · The architecture that survives
The exit is written first, and every choice at acquisition is judged against it.
What the constraints leave standing is not a preference between the four routes. It is a rule about the order the design is done in, and a structure that keeps more than one route alive for as long as the fund holds the asset.
The exit test is fixed at entry, in one paragraph, and every structural choice at acquisition is judged against it: how the capital comes home, on what date, and under whose definition of done.
The entity that will be sold is identified at formation and built for that purpose: resident where the design requires, holding the asset and nothing else, contracting on terms a share buyer would accept, and audited from the first year. The consent map is drawn at entry, listing every party whose agreement the exit will require and what each of them will want at the moment they are asked. Where the buyer universe is genuinely small, the structure keeps two routes open rather than one, because a single-route exit is priced by the only buyer who knows it is the only buyer.
Where none of this was done at entry and the term is already short, the honest answer is that the design work is now a choice between a continuation and a sale on somebody else's terms, and that choice is made with a full account of what each one costs the register rather than with an assumption that one of them is a formality.
What the one paragraph has to contain
It names the route the design is built for and the route kept alive beside it. It names the party that has to be willing to buy, in categories rather than in names. It names the date by which the capital has to be back on the register, taken from the term rather than from an intention. And it names who decides that the exit is done, because a partial realisation, a rolled interest and a cash payment are three different answers to that question and the register does not treat them as one.
Keeping the second route open is not free. An entity maintained for a share sale carries its own accounts, its own board, its own filings and its own annual test. What that buys is the removal of the fact that makes the last conversation expensive.
A structure with one exit route is not a structure with a risk in it. It is a structure whose price at the end is set by somebody else, at the moment the term becomes visible.
07 · The implementation framework
Six items in order, and the signature on every one of them belongs elsewhere.
Written as dependency statements rather than as tasks, because the sequence is the part that is easy to get wrong and expensive to correct.
Write the exit test before the acquisition structure is drawn. Draw the consent map from the contracts. Identify the entity that will be sold and build it to be sold. Fix the valuation arrangement at entry, not at exit. Diarise the annual test that keeps the second route alive, and give it an owner inside the manager rather than a place in a plan.
The signatures then belong elsewhere, as they should. The manager's counsel reads the transfer, pre-emption and change-of-control provisions and states what each one requires. Its tax advisers take the treatment of a share sale against an asset sale, for the fund and for the holders, because the two are different transactions with different answers. The valuer and the process party are appointed by the manager and answer to their own standards. The lender answers the change-of-control question under its facility. Our work is the exit test, the consent map, the comparison of the four routes and the annual test that keeps more than one of them open.
The sequence, and what each step depends on
- 01 The exit test, written Depends on nothing except the manager stating what it wants the end to look like. The only step with no precondition, and the one most often taken last.
- 02 The consent map, drawn from the documents Depends on the asset contracts, the facility and the constitution being in one place. Counsel states what each provision requires; the map states who has to be asked, and when.
- 03 The holding form, chosen against the test Depends on both of the above, and is the one-way step. Where the entity route is taken, the entity is specified now: what it holds, where it is resident, and what it may sign.
- 04 The valuation arrangement, fixed Depends on the manager appointing before it needs a number. Fixed at entry the arrangement is a control. Fixed at exit it is an argument.
- 05 The annual test, diarised and owned Depends on a named person inside the manager rather than on a plan. Four questions, one date in the year, a written answer that is kept. The only item that runs for the whole hold.
- 06 The handover to the parties who sign Depends on the five above being on one page. Each party then receives a question that is already the right question for them.
The order is the whole of it. Each of these six is cheap in the sequence written above and expensive in any other.
08 · What this case generalises to
Wherever the buyer can be counted at entry, the structure is designed backwards or not at all.
Four families of transaction share the shape of this one. In each, a number that is small and knowable at the start is treated as a problem for later, and the party who eventually reads it is not the party who created it.
Land where foreign ownership is restricted, licensed operating businesses, interests that carry a permission with them. The ownership rule sets the buyer universe before any commercial question is asked. The structural response is to reach the wider universe through what may be sold rather than through what may be owned.
A closed-ended vehicle holding a position that does not naturally realise inside the term, whether that position is an infrastructure interest, a development, a credit position with a long tail, or a stake in a business with one plausible acquirer. The date is a fund fact and the realisation is an asset fact, and the two are reconciled by design or by discount.
A residence position, a clean entity, an unbroken audit history, a licence maintained. Each is cheap to hold and impossible to restore, and nothing in the ordinary running of a fund prompts anybody to check it. One owner, one date in the year, a written answer that is kept.
A continuation, a transfer between two vehicles of the same house, a distribution in specie. The structural question is never the number. It is whether the number was produced by a process that can be shown to the people entitled to disagree with it, and that is settled by appointments made before anybody needed a number.
The engagement that answers this class is a Structural Risk Review: the exit test, the consent map, the comparison of the four routes to the same depth, and the annual test that keeps more than one of them open. It stops where the signatures start.
Every exit route but one is closed at entry, and it is closed by people who are not thinking about the exit. The cost of that is paid once, by everybody, at the end.
Written as a type · no party, no value and no date · stated as at August 2026
Write to us.
Complex transactions fail at the interfaces between otherwise workable components. We resolve the structural complexity between investment intent and transaction execution.
Write to us