Most of what makes a system thrive is relational — and it is the one kind of wealth with no instrument to invest in it. Where capital should flow, and why it doesn’t. The Great Homecoming research programme · June 2026
We allocate capital toward what we can see and price: buildings, equipment, balance sheets — the material layer of a system. But by the World Bank’s own accounting, most of what actually makes a country, a company or a community work is not material at all. It is relational and institutional — the rule of law, the skills people carry, the trust and working bonds that hold the parts together. That layer is the dominant share of wealth, and it is the only major asset class with no investment instrument attached. The consequence is a systematic misallocation: money pours into the visible layer while the returns concentrate in the relational one. This article makes the structural case that the highest-leverage investment is usually in the bonds, not the nodes — and is honest about which parts of that claim are measured, which are literature-supported, and which are, for now, only modelled.
Ask where investment goes and the answer is legible: plant, property, infrastructure, inventory, financial instruments — things that can be counted, owned and sold. This is the material layer of a system, and it has a deep, sophisticated machinery for pricing and funding it. None of that machinery is wrong. It is simply pointed at one layer of a system, and not the one that does most of the work.
A system is not only its parts; it is the bonds between them — the relationships, loyalties, shared purpose and working trust that turn a collection of parts into something that holds. A hospital is not its building; it is the coordination among the people in it. A market is not its firms; it is the institutions and trust that let strangers transact. The bonds are the integrating layer — and they are almost invisible on a balance sheet.
This is not a metaphor — it shows up in the measurement. The World Bank’s own accounting of national wealth finds that most of it, roughly three-quarters worldwide, is intangible: not buildings, machines or natural resources, but things like the rule of law, education, and the trust that lets people cooperate. Physical capital is the small slice; the relational and institutional layer is most of the pie. (These are shares of accumulated wealth, not the return on the next dollar — a distinction that matters, and one we come back to.)
So the relational layer is not a soft add-on to “real” wealth. By the most mainstream measure there is, it is most of the wealth. And it is the one major part of it with no way to invest: there is no asset you can buy that is a stake in a society’s trust, a team’s coherence, or an institution’s capacity to hold itself together. The largest form of wealth is the one with no market — and that gap is what this article is about.
There is a clear reason the returns should concentrate there, and a development record that shows they do.
The reason is that a system’s output is multiplicative: it runs only as well as its weakest necessary part, so value concentrates on whatever is scarcest. Add more of a factor that is already plentiful while the scarce one stays starved, and you buy almost nothing. When the scarce factor is relational capacity — coordination, trust, working institutions — pouring in material capital is pushing on the part that isn’t the constraint.
The record bears this out. Where institutions are weak, a dollar of public investment has been found to create about fifty cents of real capital, or less; in some settings the return on extra physical investment runs close to zero, because the binding constraint was never money but the capacity to use it well. The same project, run under sound institutions and under distorted ones, returns markedly less in the second case. The pattern is consistent: material investment laid over a thin relational layer leaks away.
So the advantage of investing in the relational layer over the material one is real — but it is conditional, not fixed. Across the evidence it runs from roughly two-fold to perhaps fifty-fold: largest where relational capacity is most depleted, and shrinking toward no advantage at all where that capacity is already strong. This is the practical heart of the thesis. The relational layer is not always the better buy; it is dramatically the better buy where the bonds are thin. The number tells you where to invest, not only what in.
There is a sharper version of the same finding inside the framework. A system sits inside a larger one, and what holds it together is its holding layer — the binding that makes the whole more than the sum of its nodes. The structural claim is that repairing this holding layer moves far more than patching the individual parts beneath it: fix the coordination, the trust, the shared frame, and every node it connects works better at once; fix one node while the binding stays broken and the gain stays trapped in that node.
How much more? In our own model simulations, repairing the holding layer moves roughly thirty times what the same effort spends on individual parts. That figure comes from the model, not from the world — a simulation result, not externally validated — so we treat it as illustrative, not as a measurement.
What we can say with more confidence is the direction and shape: investment in the relational layer dominates where that layer is weakest. And unlike the headline number, that claim is testable. It predicts a specific ordering — the biggest relational premium in the most depleted settings, the smallest in the healthiest — which real deployments can confirm or break. If the ordering fails outside the model, the theory is wrong, and we have said so in advance.
If the relational layer is where the returns are, the obvious question is how to invest in it deliberately — and that is exactly what an instrument like Flowcoin — the allocation model the programme is developing, which funds verified outcomes rather than promises and is itself still under construction — is designed to do. Paying for relational results is harder than paying for buildings, for two reasons, and the design answers both head-on. Verifying each result can be costly, because relational outcomes are small and frequent; and it is easy to pay for false credit — change that looks delivered but either didn’t happen or would have happened anyway. (In carbon-offset markets, an estimated nine in ten audited credits turned out worthless for precisely this reason: not bad measurement, but inflated baselines.) The answer is to recognise value only when two independent checks both hold.
The first is Proof of Benefit: was integration actually delivered? Read not from a self-report but as a measured improvement in the system’s state, against a baseline taken from independent records, net of any disorder the project pushed onto others. This is how much good was really done.
The second is Proof of Orientation: which way is the project pointed? Real benefit can be delivered in the service of the wrong end — capable good works that entrench capture — so benefit alone is not enough. Orientation reads where a system is actually aimed, disciplined by the gap between what it declared it was for and what it did: the say-do gap. Value is recognised only when both hold — genuine benefit, honestly pointed.
These are not vague virtues; they are quantities the framework already measures — the orientation of a system, the charter it declares, and the divergence (σ) between that charter and its conduct. An instrument built on them does something an ordinary funder cannot: it turns every funding decision into an explicit forecast, then scores that forecast against what the money actually achieved — so it keeps an open record of its own accuracy and gets sharper the longer it runs. A small share of funding is placed deliberately outside its own preferences, so it keeps learning where its judgement is weak rather than only confirming where it is strong. And because the scarce factor is usually relational capacity, the highest-return move is often to build that capacity first — investing in people before the projects that will need them.
Put together, this is a new kind of instrument: a model of fund allocation aimed at the layer where most of the world’s wealth actually sits, and where the returns are highest precisely because almost no one invests there. It pays only for benefit that was really delivered and honestly aimed, and it improves at choosing the longer it operates. The potential is large. Most development and philanthropic capital today is poured into the visible material layer and leaks away against a thin relational one; an instrument that redirected even part of it toward the scarce, high-leverage factor would do an unusual amount of good per dollar — and, unlike the asset class it targets, it would carry its own evidence with it.
It is, today, a design backed by model results rather than a finished, externally tested instrument — the decisive proof comes in deployment, against systems the model has never seen, and that work is what lies ahead. But the case for building it is strong, and what it points at is the largest under-served asset class there is.
The Great Homecoming is an independent research programme on why systems cohere or fragment. Companion pieces: “The Limits of the Market” — the inversion that happens when a good is priced purely as an asset; “Choosing Leaders Who Can Integrate” — the holding layer at the level of leadership; “The Success That Stopped Growing” — reading the bonds before they break. Contact: Wim Van Laere.