The real bottleneck in credit is operational.

There are good borrowers, good originators, and available capital. The problem is making it all talk to each other.

I have spent much of the last few years trying to understand where money gets stuck. First I thought the problem was simply finding capital. Then I saw there was a lot of money in the world and few good places to put it. I began paying attention to origination, to selecting borrowers, to the quality of collateral, to tenor, to structure, and to the relationships that let a private operation appear before it becomes merely one more product on a shelf.

Today I am beginning to suspect there is a more banal difficulty cutting across all the others.

The parts exist.

They simply do not talk to each other well.

That looks like a small conclusion next to the great discussions about interest rates, liquidity, monetary policy, and credit risk. Perhaps that is precisely why it took me so long to see it. We are naturally drawn to intellectually prestigious problems. It is far more interesting to discuss a company's cost of capital than to admit that an operation worth a few million may be stalled because nobody knows which is the latest version of the balance sheet.

Reality has that unpleasant habit of humiliating abstractions through administrative details.

Over the last few years working as a loan broker, I have followed the whole process much more closely. The businessman arrives with a need. Sometimes he already knows exactly how much he needs. Often he believes he knows. We start looking at the business, the flows, the estate, the debts, the collateral, the purpose of the funds, and what actually has to happen for the capital to solve the problem instead of simply postponing it.

Then we have to turn that reality into something comprehensible to whoever has money.

That is where the operation starts fragmenting.

The client talks to the accountant. The accountant sends documents. Some version is out of date. The bank requests others. Legal has to verify a piece of collateral. A second institution wants the same information in a different format. A fund asks for something the previous bank did not. The businessman does not know whether a certain file was already forwarded. I have to find out which institution received what, when it received it, what it answered, and what the next outstanding item is.

There may be an excellent company at the center of all that.

There may be a creditor willing to lend.

Even so, the money does not move.

I am beginning to think we call "the credit market" something that, observed closely, still depends extraordinarily on manual work to coordinate information among people using different systems and rarely sharing context.

Perhaps the real problem is not at the ends of the operation.

It is in the middle.

That observation gained an interesting coincidence this week. The Central Bank reported that Brazil's Open Finance has already reached about fifteen million unique clients, tens of millions of consents, and more than eight hundred participating institutions. The number is impressive, but what really interests me is in the technical explanation: APIs let systems from different institutions talk to each other securely, quickly, and in a standardized way.

"Talk" is a simple word for an important change.

For decades, much of the financial system worked through islands. Each bank knows its own part of the client. Each institution keeps its own records. The client has a financial reality and each supplier has only a slice of it. To move the relationship, it was frequently necessary to reconstruct information, resend documents, and accept that the new institution would start practically from zero.

Open Finance tries to change part of that architecture.

The client authorizes.

The data can circulate.

Systems stop depending exclusively on human beings copying information from one interface to another.

It is early to know how transformative that will be. Every piece of financial infrastructure is presented initially as a revolution; then we discover that people are still people and organizations are still excellent at preserving old incentives using new technology. Even so, the principle seems much larger to me than any specific feature.

When systems start talking, economic value migrates.

Owning the information stops being enough.

One has to know what to do with it.

That brings me back to origination.

A few years ago I imagined a good loan broker had to build a large network of institutions. I still believe that. There is no true independence when the intermediary knows only one source of capital. If he works with a single bank, sooner or later he will start finding that many problems curiously have exactly the solution that bank sells.

But knowing institutions is not enough.

A network without process starts becoming a problem as it grows.

I can know twenty sources of capital. That widens my options. It also means I have to know what each one is looking for, which operations they are analyzing, which documents they require, which proposals they made, and why they refused the last deals they received.

Richness of relationships produces operational poverty when there is no infrastructure to organize it.

It is a rather common paradox in intermediation businesses.

The professional starts small and his head works perfectly as a system. He knows five clients, a few institutions, and a handful of simultaneous operations. He knows everything because everything fits in memory.

Then he starts getting good.

More clients appear.

Other originators start bringing opportunities.

Institutions start answering.

The volume grows.

And precisely what seemed to demonstrate the business is working starts degrading its quality.

The professional spends more time administering what he originated and less time doing what he actually has an advantage in.

The demand itself consumes his capacity.

It is a kind of intellectual working capital.

Growth requires structure before it returns scale.

I have thought a lot about that analogy because it explains why many service businesses seem to reach a plateau. The founder is excellent. The market wants more of him. So he hires people to do what he was doing. But part of the value lay precisely in the context that individual carried. In trying to scale human work through more people, he risks diluting what made him valuable.

There are two possible solutions. The first is never to grow much.

That can be an excellent choice.

There is no moral obligation to turn every good business into an empire. Some activities perhaps should stay small because proximity is part of the product.

The second is to discover which parts really have to stay human and build infrastructure so the others stop consuming the human.

That second option is beginning to interest me.

In loan broking, judgment has to stay human.

So do relationships.

Negotiation will probably stay deeply human for a long time.

Understanding that a given businessman is telling a coherent story, sensing that a bank is interested but has not yet found the right structure internally, knowing when to push and when to wait, recognizing that a given family may understand a piece of collateral better than a generic fund: all of that depends on context.

Now, asking for a document for the third time does not require genius.

Neither does tracking which version was received.

Recording that a given institution refused an operation because its maximum LTV is different does not require continuous judgment.

Remembering that the proposal expires on Friday is not a financial activity.

It is outsourceable memory.

There is an enormous quantity of tasks mixed in with intellectual work because historically we executed everything through the same people.

Perhaps that is a bad design.

If I hired an excellent surgeon, I would hardly consider it a good use of his time to have him organize the instruments, fill in every administrative form, and phone each patient to confirm the appointment. Not because those tasks are undignified. Because we are wasting scarce competence on what can be organized another way.

Loan broking has something similar.

The scarce resource is not the email.

It is the judgment of whoever knows which email should be sent to whom.

That difference looks small until we start multiplying it by dozens of operations.

Perhaps the entire market loses an enormous amount of origination capacity because good originators spend a considerable part of the week functioning as document administrators.

That is especially contradictory at a moment when the financial system itself is starting to build interfaces able to share data automatically.

The institution is going digital.

Origination remains handmade.

There is a lag.

Perhaps it is temporary.

Someone will probably solve it.

I do not yet know what form that will take, but the problem is clear enough that I have started observing it differently.

A credit operation has a kind of life of its own. It is born as a request. It receives documents. It is framed. It goes to an institution. It comes back with questions. It may be sent to another. A proposal appears. Terms are negotiated. Collateral has to be verified. The client decides. The operation closes or dies.

All of that produces information.

Today much of it disappears inside the process itself.

If an institution refused because it does not like that sector, we should learn.

If another accepted a certain piece of collateral, that should improve the next origination.

If a client sent a certain document, he should not have to reconstruct it from scratch in every conversation.

If an operation took forty days because a certain stage sat still for twelve, we need to know.

A firm that closes operations without learning from them is converting work into revenue, but not necessarily into intelligence.

That is waste.

Experience should compound.

Every deal should make the next operation slightly better.

The network should gain memory.

That may be the point at which information stops being a file and becomes infrastructure.

I also think about what that means for a family office.

For some years I have been arriving, by different routes, at the conclusion that a family office's real work is far larger than choosing investments. The family has an economic life, not a portfolio.

There is an operating company.

Properties.

Stakes.

Funds.

Cash.

Debts.

Collateral.

Corporate structures.

Banking relationships.

Future commitments.

Perhaps investments abroad.

Private operations.

Club deals.

Children who will take part in the next phase.

Decisions made years ago for reasons nobody may have recorded.

The more the wealth grows, the more context has to be preserved.

That is precisely why the concept of a multi family office strikes me as so interesting and so dangerous at the same time.

Interesting because several families can share infrastructure that would be economically absurd for each to build alone.

Dangerous because scale can destroy context.

A single family office serves one family. It is possible that certain important information exists only in the memory of a few people, though that is also a risk.

In an MFO, that informality stops working quickly. Ten families do not mean ten portfolios. They mean ten histories, dozens of companies, hundreds of relationships, and thousands of documents and decisions.

Without serious infrastructure, the office starts choosing between two bad things: depending on the memory of overloaded people or turning distinct families into standardized records.

The first does not scale.

The second scales while destroying part of the value.

Perhaps the real challenge is achieving structure without losing context.

That is harder than it looks.

Software likes fields.

Families have exceptions.

The system asks for "value of the property."

The family answers that the property is worth twenty million and will never be sold because the grandfather built the house.

Financially there is an asset.

Patrimonially there is something else.

The system can record the number and still lose the reality.

The same happens with companies.

A stake can be worth a hundred million and be practically illiquid. It can pay dividends. It can be the whole family's main source of income. It can be the center of its identity and of its political and social relationships.

Reducing all of that to "equity: R$100 million" organizes data and destroys context.

Technology that serves wealth has to know that not every important piece of information is a cell.

That may be one of the reasons human intelligence remains central.

The system should carry the memory.

The person should interpret.

That separation seems more and more promising to me.

A good family office could know, at any moment, how much the family owns, where it owns it, how much it owes, which collateral is committed, which private investments will require future capital, which documents expire, which banks are involved, and where risk is concentrated.

That looks basic.

Perhaps precisely because it is basic it is so rare.

Then, on top of that infrastructure, the adviser can exercise judgment.

Not the other way around.

Today we frequently do something rather peculiar: we use human intelligence to reconstruct facts a system should know and then little time is left to think about what only humans can decide.

It is a poor distribution of competences.

I also think about how that infrastructure could change credit for business families.

A businessman goes to an institution and has to prove his wealth.

The bank asks for information.

The client sends documents.

A year later he approaches another institution and starts over.

If he has a family office able to keep the consolidated view, the dynamic changes. The client is not improvising a patrimonial photograph for a specific operation. A permanent financial memory exists.

That can reduce the cost of capital.

Not because the family became richer.

Because it became more legible.

Legibility is worth money.

The creditor applies a premium to what it does not understand.

If the wealth structure is fragmented, the institution has to assume there are things it cannot see. That can mean a lower limit, more collateral, or a worse rate.

Organization reduces operational uncertainty.

It does not eliminate economic risk.

But it may allow economic risk to be priced with less noise.

There is a difference.

A bad company does not become good because it has pretty documents.

A good company can certainly look worse when nobody can explain its finances.

The same holds for a family.

That makes the back office look like a direct component of wealth creation, something I would not have imagined a few years ago.

Better information can produce better credit.

Better credit reduces cost.

Reducing cost increases wealth.

The chain is not glamorous.

It works.

Perhaps there is a human tendency to look for additional return in exciting places before removing visible waste. The investor looks for a fund yielding two extra points and ignores a debt costing four points more than necessary because nobody consolidated the two pieces of information.

It is a kind of unrealized internal arbitrage.

The estate has capital invested on one side and expensive capital borrowed on the other.

Two institutions profit.

The family does not notice because each relationship sits in its own drawer.

A true family office should see those contradictions.

That may be one of its greatest advantages over specialized suppliers.

The manager tries to improve assets.

The lender thinks about liabilities.

The lawyer thinks about structures.

The family office should be able to put all of it on the same table and ask whether the whole makes sense.

That looks more and more like coordination than like management.

Coordination is not a prestigious word.

Nobody dreams as a child of "coordinating capital."

Perhaps that is why the work is badly served.

The market sells products because products have a clear price and commission. Coordination produces value through relations among things. It is harder to demonstrate and therefore harder to charge for.

A multi family office could change that if it can turn coordination into a product.

Not a financial product.

A continuous service.

The family pays so that its financial life stays organized, comprehensible, and ready for decisions.

Investments still exist.

Credit.

Structuring.

Club deals.

Private markets.

But all of them sit on a base.

That logic seems particularly important in the private market because there is not as much ready-made infrastructure there as in the public one.

When I buy a listed share, someone has already solved custody, registration, pricing, liquidity, periodic information, and a series of mechanisms I rarely have to think about.

When a family lends directly to a company, almost everything has to be created for that relationship.

Origination.

Underwriting.

The contract.

Collateral.

Monitoring.

Collection.

Eventual renegotiation.

If it does a club deal with other families, it adds governance among the investors themselves.

The return can be better precisely because the private market is less organized.

That means part of the spread pays for the disorganization.

The investor who can build infrastructure to reduce that friction can capture a real advantage.

But he should not imagine the whole spread is alpha.

Part of it is a disguised back office salary.

That thought may help explain why some private investments look excellent in the presentation and far less extraordinary once we account for the real cost of operating.

A family office can earn 15% on a direct credit.

Excellent.

How many hours were spent originating, analyzing, documenting, monitoring, and collecting?

What concentration risk?

What liquidity?

What would it cost to reproduce the process?

If all of that is ignored because the capital already belongs to the family, we are mixing financial return with unaccounted work.

That does not make the operation bad.

It makes the comparison more honest.

Perhaps an MFO again has an advantage because it can spread part of that infrastructure across several families and operations.

An analyst can follow a larger portfolio.

A system can organize several deals.

Relationships with originators are reused.

Lawyers know the patterns.

Experience accumulates.

The operational cost per operation falls.

It is similar to what a fund does, but preserving each family's ability to decide its own exposures.

That combination is intellectually attractive.

Economies of scale without necessarily putting everyone inside the same product.

There may be regulatory, alignment, governance, and conflict problems. There certainly will be.

But the mechanism is interesting.

The asset is not merely access to a club deal.

It is the infrastructure that lets you assess and monitor many club deals without turning each into a small administrative company.

Access without operational capacity is a form of excess.

That may also be true of my own network.

The more originators, businessmen, and institutions I know, the greater the potential quantity of deals.

It would be easy to treat that growth as an absolute advantage.

It is not.

Too much deal flow can worsen decisions if the capacity to filter does not grow with it.

A broad network requires an even better sieve.

Otherwise the good originator becomes a mail carrier.

He receives.

He forwards.

He waits.

He chases.

Perhaps the path to professionalizing the activity is removing exactly that logic from it.

The client should not have to depend on messages asking "any news?"

It should be possible to know what state the operation is in.

The broker should not have to depend on his own memory to find out who needs a follow-up.

The information should exist independently of anyone remembering it.

That looks small, but it changes the relationship.

When the process is opaque, the client has to chase the intermediary.

When it is legible, the intermediary can spend energy solving the problem.

Perhaps operational transparency is a form of trust.

Not the personal trust built over years, which remains fundamental.

A second layer.

I trust because I can see.

There is a lesson there for family offices too.

Wealth is usually administered through a relationship of high trust and low visibility. The family trusts the adviser, receives periodic reports, and asks when it needs something.

Perhaps that model will inevitably be changed by technology.

The client will start expecting a more continuous view.

Not because he wants to operate alone.

Perhaps the opposite.

He wants to delegate without going blind.

Delegation and opacity do not have to be synonyms.

An excellent autopilot does not require the passenger to ignore where the plane is going.

That may be another characteristic of good infrastructure: it increases delegation because it increases trust in what is being delegated.

I do not want the client to have to take part in every document of a credit.

I want him to know the state.

He does not have to analyze every financial position.

He has to know the consolidated view exists.

He does not have to negotiate every service provider.

He has to know who was hired and why.

That clarity produces a different relationship.

Perhaps even more important for families with more than one generation.

The founder can trust the adviser personally because he has known him for twenty years. The children may not.

If the service depends exclusively on the relationship between two men, the family office faces a succession problem as large as the family's own.

The next generation has to be able to trust the institution, not merely the father's friendship.

That requires process.

Memory.

Documentation.

Transparency.

The human relationship continues.

But now there is something able to survive the people.

Perhaps that is what separates an institution from an excellent professional.

The professional takes his intelligence with him.

The institution turns part of the intelligence into shared memory without making anyone irrelevant.

It is a difficult balance.

It may be exactly what I am trying to understand when I look at my own work as a loan broker.

Today a relevant part of the value is still inside my head.

I know clients.

Sources.

Policies.

Previous operations.

That is good.

It is also a risk.

If I want a business that exists beyond my individual capacity, I have to make part of that memory belong to the business without the business starting to act like a bureaucratic robot.

Process should increase judgment.

Not replace it.

That may be the rule.

There is a frequent temptation to solve disorganization by adding bureaucracy. The process becomes a collection of steps nobody knows the reason for. Then we hire more people to administer the steps.

That does not interest me.

A good process eliminates work.

If a piece of information already exists, do not ask again.

If an institution has no appetite for a given structure, do not send it.

If a document is not necessary at that stage, do not require it.

If the client has to answer for something, make clear why.

Perhaps operational excellence is simply respect for people's time.

There are firms that confuse friction with rigor.

They are not the same thing.

A process can be rigorous and fast if the information is right.

It can be slow and incompetent without producing any reduction of risk.

Banks know that problem well. Some requests exist because they are necessary. Others because nobody took the risk of removing them.

Bureaucracy has a political advantage: if something goes wrong, someone can say he followed the process.

That is very different from the process having improved the decision.

Perhaps good infrastructure for origination has to begin each step by asking what problem it solves.

If we do not know, perhaps it should disappear.

That mentality looks obvious in a startup.

It is surprisingly rare in financial services.

Perhaps because the cost of inefficiency is passed on to the client.

As long as the businessman keeps sending files, the system survives.

That can change as competition increases.

Open Finance may be one piece of that movement.

If data starts circulating, institutions able to use it with less friction can win clients.

The portability of information reduces the power to hide bad processes behind switching costs.

It is early.

But I like the direction.

The client should be able to carry his financial context.

The more portable the information, the less dependent he is on a specific institution.

That strengthens precisely the role I have been imagining for an independent adviser or family office.

The bank keeps one part.

The family office keeps the view.

If the bank stops being competitive, you change banks.

The context remains.

If a lender no longer makes sense, you look for another.

If a manager loses quality, you replace him.

The estate should not be reconstructed every time a supplier changes.

That may be the correct architecture.

The client at the center.

Suppliers around.

Data organized enough that switching does not destroy memory.

A trusted institution coordinating.

I do not know whether the financial system will move exactly in that direction.

But it seems economically more rational than forcing every client to reconstruct his own history before every new supplier.

There is something almost medieval about the current model.

The banker holds the record.

The client asks permission to take his own information.

Open Finance, at least conceptually, starts inverting that.

Perhaps the next step is letting the client have his own layer of intelligence over that data.

Whoever can build that may occupy an extraordinary position.

He does not have to manufacture every financial product.

He can simply understand the client better than whoever manufactures them.

That is a big idea.

It is also dangerous.

Whoever knows a person's or family's entire financial life receives enormous power.

Knowing wealth, debts, cash flow, investments, and relationships creates the capacity to recommend.

It also creates the capacity to exploit.

Technology does not solve ethics.

It may increase the need for it.

An adviser with a partial view can do partial damage.

An institution with the complete view can make far better decisions and, if misaligned, extract far more.

That is why the compensation model and the incentives will remain fundamental.

If the institution that organizes the estate earns more when the client buys a given product, the consolidated intelligence can turn into an extraordinarily efficient distribution machine.

That would be exactly the opposite of what I want.

The infrastructure should increase the client's autonomy, not the intermediary's capacity to capture him.

That principle will have to be preserved if I ever build something in that direction.

I do not yet know what.

But I know what I do not want.

I do not want a shelf disguised as advice.

I do not want a machine that learns everything about the client merely to discover the next product it can sell.

I want the context to allow choosing better, including choosing not to buy.

That holds for credit.

A system that knows the businessman should not use the data merely to increase loan volume.

It should be able to see when credit is unnecessary.

Or when the current debt has to be reorganized before another is taken on.

Perhaps true financial intelligence sometimes produces fewer transactions.

The market has difficulty monetizing that.

It is an interesting problem.

If a firm earns per operation, commercial efficiency means more operations.

If it earns on the quality of a continuous relationship, it may be economically rational to reduce bad transactions.

That second model looks closer to what a family office should do.

It may also be the path to professionalizing loan broking in a less conflicted way.

Not merely selling capital.

Charging for the infrastructure, the organization, or the advice that lets the client access capital when needed.

It is a hypothesis.

I do not yet have an answer.

But I notice my discomfort with the transactional model is growing.

There is a limit to the quality of advice when the only revenue appears if the advice ends in a contract.

The lawyer is paid to advise even when the recommendation is not to sue.

The doctor is paid even when he concludes that surgery is unnecessary.

Why is so much financial advice paid for only when something is sold?

Perhaps because a product is easy to charge for and judgment is hard to price.

But that ease may be costing alignment.

A multi family office could solve part of that by charging for the relationship and using the financial market as a supplier.

I do not know whether every family will accept paying explicitly for something they now believe they receive "for free" from banks and brokerages.

Nothing is free.

When the price does not appear, it is usually hidden somewhere harder to compare.

Perhaps financial maturity is, among other things, preferring visible costs to invisible incentives.

It is an idea I want to keep testing.

For now, 2023 is leaving me with a simpler conviction.

The problem is not that pieces are missing.

There is capital.

There are businessmen.

There are investors.

There are family offices.

There are banks.

There are funds.

There are originators.

There are documents.

There is data.

There is technology.

There are lawyers.

There are managers.

Practically everything exists.

The inefficiency appears because each piece keeps its own view of the world and much of the coordination is still done manually by people trying to remember what the other party needs.

Perhaps the real bottleneck is exactly there.

Interoperability.

Not merely among APIs.

Among people, processes, and incentives.

A system can make two databases talk.

It is harder to make two institutions understand the same operation the same way.

It is harder still to preserve the client's context while that happens.

That may be the layer that still has to be built.

A kind of infrastructure of the middle.

Not the capital.

Not the client.

What connects, organizes, records, and gives continuity.

It is curious to notice I have spent fifteen years interested precisely in the intermediate position.

First as a salesman.

Then looking at money.

Then trying to understand intermediation.

Loan broking.

Family offices.

Now I am beginning to see that the greatest value of intermediation may not be merely bringing two parties together.

It is building the infrastructure through which they can work together repeatedly.

The introduction produces an operation.

The infrastructure produces a market.

That difference may be far larger than I understood.

A broker can close ten deals.

An infrastructure lets dozens of brokers, clients, and sources of capital organize hundreds.

A family office can know one family.

Adequate infrastructure may let a multi family office preserve the context of many families without turning them into numbers.

I do not yet know whether those things belong to the same problem.

I suspect they do.

Both involve money scattered among systems that see only fragments.

Both depend on trust.

Both have repetitive operational work.

Both need memory.

Both suffer when they grow.

Perhaps there is something at the crossing.

I do not have to name it now.

I have learned that premature names are dangerous because they make a person start defending a solution before understanding the problem.

I would rather keep observing.

But the question has changed.

It is no longer merely how to sell money.

Nor merely how to find capital.

Nor how to structure a good operation.

The question beginning to interest me is another:

how do you build a machine in which all of that can happen without depending on improvisation?

If I figure that out, I may have found something larger than a profession.

Leo Bentier