Bringing the Money With You

Volume I — Transition: From Here to the Living Economy

THE PLATE

A woman at a broad wooden desk sorting papers, afternoon sun pouring through the tall windows behind her. — Plate I.06The Stack, in Afternoon Light.Capital is never one thing. It is three or four things placed in an order, and the order is the whole design. The stack on the table is not money yet — it is the shape the money will take.

THE LETTER

You have a result. Chapter I.05 gave you a pilot that paid, and a signed baseline behind it that nobody can argue with. The question that arrives next is the one nobody warns you about, and it arrives faster than you expect: where does the money come from when the thing is ten times bigger?

The instinct is to go back to the same place the pilot came from — the discretionary line, the unspent budget, the one persuaded signature. That works once. At ten times the size it does not, because you have left the territory where a single person's authority reaches, and you have entered the territory where capital has to be raised on terms and priced against alternatives.

Here is the sentence this chapter exists to deliver: the money already exists, it has names, and it has published terms. Not a fund that might be persuaded to exist. Not a mandate that will appear if enough people ask. Revolving funds with fifty years of loss history. Community development lenders whose sector net charge-off rate is 0.58 percent. Blended structures with a documented four dollars of private capital for every dollar of concessional. Guarantee programmes that will take eighty percent of a credit risk for a fee you can compute on one line. Patient equity that accepts two percent where Treasuries pay four, and publishes the offering memorandum that says so.

None of that is philanthropy asking to be found. It is an asset class with a market, a pricing convention and a set of counterparties who take meetings.

What this chapter gives you is the stack: how the four kinds of money fit together, what each one costs, what each one asks for in exchange, and the one arithmetic that tells you whether your version of it holds. It also gives you the case where this loses, because there is one, it is common, and it is expensive — an instrument thousands of firms have signed where the price benefit is genuinely smaller than the cost of proving you earned it.

You are not going to become a fundraiser. You are going to learn to read four term sheets and put them in the right order.

— The Editors

DISCOVERY

The capital that is already doing this, and what it actually returned

Begin with what has already been built, because it is more than most people working in this field believe, and the terms are public.

Self-Help, Fannie Mae and the Ford Foundation. In 1998 the Ford Foundation placed $50 million with the Center for Community Self-Help in Durham, North Carolina. It was not a grant to make loans. It was a grant to sit behind loans — a credit enhancement, so that Fannie Mae could buy mortgages it would otherwise have declined, originated by conventional lenders to borrowers with thin files and irregular income. The Community Advantage Program that resulted financed more than $4.5 billion of mortgages for over 50,000 families across 48 states. Fifty million dollars of philanthropic capital standing behind four and a half billion of private capital is a leverage of ninety times, or about one thousand philanthropic dollars per family housed. The loans performed: the UNC Center for Community Capital followed them through the foreclosure crisis and found that these borrowers kept paying while comparable subprime borrowers did not.

The Clean Water State Revolving Fund. Begun in 1987, the CWSRF is the least glamorous and most instructive piece of regenerative finance in the United States. The federal government capitalises a fund in each state; the state matches twenty percent; the fund lends to water infrastructure at below-market rates; borrowers repay; the money is lent again. Cumulative federal capitalisation had passed $42 billion by 2017, and cumulative assistance passed $153 billion by 2021 — 3.6 times the federal contribution, and just over three times federal and state money together. About a third of what goes out is leveraged bond proceeds; the rest is repayment, recycled. The mechanism has no ideology in it at all. It is a stock with a regeneration rate, managed as one.

Iroquois Valley Farmland REIT. A public benefit corporation that buys farmland and leases it to organic and transitioning farmers on long leases, and funds itself partly through Soil Restoration Notes — promissory notes with a coupon between zero and 2.50 percent, in three, five and seven-year terms, sold under Regulation A to ordinary investors. More than $115 million has gone into organic farmland this way: about 32,000 acres, 70 farms, 20 states, and half a percent of the raise is set aside in a pool that reimburses tenant farmers for soil amendments, agronomy and seed during the transition years when yields dip and certification has not yet arrived. An investor accepting two percent when a five-year Treasury pays four is making a two-hundred-basis-point gift, on purpose, in a document filed with the SEC.

Ecuador and the Galápagos. In May 2023 Ecuador repurchased $1.6 billion of its own bonds for about $656 million — 41 cents on the dollar — and refinanced with a new blue bond wrapped in $656 million of political risk insurance from the US International Development Finance Corporation and an $85 million guarantee from the Inter-American Development Bank. The enhancement is what made the new debt cheap enough for the trade to work. It produced roughly $1.1 billion of debt-service savings over seventeen years and directs about $450 million into marine conservation over eighteen, with an endowment designed to yield about $12 million a year in perpetuity. Two guarantees, one country, one ocean.

And the aggregate. Convergence, which keeps the database, records cumulative blended finance transactions of about $213 billion across more than 1,200 deals by mid-2025, mobilising on average four dollars of private capital for every dollar of concessional capital — and 4.37 when concessional guarantees are excluded from the calculation. In 2024, guarantees were 46 percent of all concessional instruments used.

Five cases, one pattern, and it is not the pattern people expect. In none of these did anyone find new money. Somebody took existing money and changed its position in the structure — moved it from the front to the back, from a grant to a reserve, from a five-year tenor to a twenty-nine-year one, from a return requirement to a return preference. The money was always there. What changed was where it stood when something went wrong.

THE ARITHMETIC

What the stack costs, what a guarantee buys, and where this loses

First, the stack, priced.

A blended facility is four layers with different appetites. Write them as weights and rates and the whole thing becomes one line of arithmetic.

  layer                      weight    cost     contributes
  ----------------------------------------------------------
  first loss / grant            10%    0.00%        0.000%
  subordinated, patient         20%    2.00%        0.400%
  senior, commercial            70%    6.00%        4.200%
  ----------------------------------------------------------
  blended cost of funds        100%                 4.60%

Against that, the book. Lend at 8.0 percent, lose 0.58 percent — the community development finance sector's own net charge-off rate — and run it for 2.5 percent of assets:

  8.00% yield − 0.58% losses − 2.50% opex  =  4.92% net
  4.92% − 4.60% blended cost               =  +32 bps

Thirty-two basis points. It clears, and it clears thinly, and that thinness is the honest reading of this entire asset class. Move the senior tranche from 6.0 to 7.5 percent — one rate cycle — and the same structure returns −73 basis points and stops. The first-loss layer is not decoration in that stack. It is 150 basis points of the 460, and it is the layer that decides.

Second — and this is the cut — what a first-loss dollar actually costs.

Everyone treats the first-loss tranche as money spent. It is not. It is money posted. Post $1 million behind a $10 million book, at the sector's 0.58 percent annual net charge-off and a three-year average life:

  expected cumulative loss   0.58% × 3 × $10,000,000  =   $174,000
  returned to the provider   $1,000,000 − $174,000    =   $826,000
  expected cost per dollar of capital mobilised       =   1.74 cents
  the same dollar given as a grant                    =   100 cents
  ----------------------------------------------------------------
  ratio                                                   57×

A philanthropic dollar placed as first loss does about fifty-seven times the work of the same dollar given away, and eighty-three percent of it comes home. That is not a rhetorical flourish; it is a loss rate divided by a tranche size, and any foundation's programme-related investment officer can check it in an afternoon.

Now hold the boundary, because a claim this large is only load-bearing if its edge is drawn. The tranche is exhausted at 5.7 times the sector's historic loss experience — a cumulative ten percent loss over three years. Below that line the senior lender is untouched. Above it, the senior lender takes the difference and the structure has failed at exactly the moment it was most needed. And two real costs never appear in the 1.74-cent figure: the first-loss provider owns the variance, not the mean, and the money is illiquid for the whole term, so it cannot also be granted meanwhile. A foundation that cannot hold a total loss of the tranche should not write it. That is the condition, and it is usually present in a foundation with an endowment and usually absent in one that spends what it raises.

Third, what a guarantee costs, in basis points.

Take a real programme. The USDA's OneRD Guaranteed Loan Program guarantees up to eighty percent of a rural business loan, for an upfront fee of 3 percent of the guaranteed portion and an annual renewal fee of 0.55 percent of the guaranteed balance outstanding. On a $4 million ten-year loan at 8 percent:

  upfront, 3% of $3.2m guaranteed              $96,000
  renewal fees over ten years                 $107,865
  ------------------------------------------------------
  total cost of the guarantee                 $203,865
  average outstanding balance               $2,451,474
  annualised cost of the guarantee               83 bps

Eighty-three basis points. And that is the wrong number to look at, because the guarantee is not buying you a spread. It is buying a yes from a lender whose credit committee was going to decline. When the alternative is a decline letter, the correct comparison is not 83 basis points against a cheaper loan — it is 83 basis points against nothing at all.

Fourth: where this loses.

The sustainability-linked loan is the instrument most firms meet first, and for a great many of them it is the one that does not pay. The structure is simple — your margin falls a few basis points if you hit agreed sustainability targets and rises if you miss. Observed ratchets run from 2.5 to 25 basis points; the academic sample illustrates a loan at 175 basis points over Euribor with 5 basis points off on target.

The break-even is one division: annual compliance cost ÷ ratchet.

  annual cost      2.5 bp        5 bp        10 bp        25 bp
  ---------------------------------------------------------------
    $40,000       $160m         $80m         $40m         $16m
    $75,000       $300m        $150m         $75m         $30m
   $150,000       $600m        $300m        $150m         $60m

(Compliance cost is a stated assumption, not a published figure — KPI data collection, limited assurance, and the legal work of the reporting schedule. Place your own and read across.)

Below the cell, the label costs you money. And there is a second, sharper term nobody puts in the model: the ratchet is earned on drawn margin, and a revolving credit facility is frequently undrawn. At a $75,000 annual cost and a 5 basis point ratchet:

  100% drawn   ->  break even at   $150m
   50% drawn   ->  break even at   $300m
   30% drawn   ->  break even at   $500m
   15% drawn   ->  break even at   $1.0bn

A mid-sized firm with a $200 million backstop facility drawn fifteen percent of the year pays the full reporting cost for a discount on a margin it is barely paying. The market has already worked this out: sustainability-linked loan deal count fell 42 percent in 2025, labelled loan volumes in the first half of 2025 were 52 percent below the first half of 2024, and borrowers dropping the label have said plainly that the benefit did not cover the operational burden.

The rule that follows is short. Below roughly $150 million of genuinely drawn debt, do not buy a margin ratchet. Buy a guarantee, a first-loss layer, or a longer tenor — all three of which pay at any size.

DREAM

What it is like when the stack is ordinary

In the economy where this is normal, the capital question is not asked as whether but as in which order.

A regional food processor putting in a water reuse system does not choose between a bank loan and a grant. It assembles: a small first-loss reserve from a community foundation whose investment committee treats reserves as investments and models their return, a subordinated note from a family office that wants five percent and twelve years rather than fifteen percent and four, and senior debt from the bank it has banked with for thirty years, which lends comfortably because it is sitting at seventy percent of a structure with a reserve underneath it. The whole stack takes eleven weeks to assemble and everyone involved has done it before.

Foundations run two books. One is the grant book, spent annually, and it is the smaller of the two. The other is a reserve book — capital posted behind lenders, priced at its expected loss rather than its face, reported on a page that shows dollars mobilised alongside dollars consumed. Programme officers know their reserve book's loss rate the way a bank knows its own, because it is published, and because after fifteen years there is a series.

Revolving funds are municipal furniture. A city has one for water, one for housing retrofit, one for small manufacturers, and nobody regards this as adventurous, any more than a sewer is adventurous. The funds grow while lending, because lending at four percent against annual losses under one percent and administration around one percent leaves the corpus larger each cycle. A dollar put in twenty years ago has been lent five times and is still there.

Tenor is matched to biology. An asset that regenerates on a twenty-year clock is financed on a twenty-year instrument, because thirty-year fixed-rate money exists for exactly this and is understood to be the point rather than an exotic option. Nobody funds a soil transition on a three-year revolver and then explains the result as a failure of soil.

And the sustainability-linked loan still exists. It is used by the fifty firms large enough for the arithmetic to work, it is priced honestly, and nobody sells it to anyone else.

DESIGN

The stack, built, in the order the money says yes

Build from the back. The most common error is to start with the senior lender, because they are the largest ticket and the easiest to find, and the answer is always a version of come back when you have the rest.

Layer one — the first loss. Go first to whoever can afford to be last. A community foundation, a family office, a corporate foundation, a state programme, a philanthropic donor-advised fund. Ask for a reserve, not a grant, and bring the 1.74-cent arithmetic with you. Size it at three to five times the expected loss on the book; ten percent of the facility is a common landing point and it should be defended with a loss history rather than a convention. Write the waterfall explicitly: what it absorbs, in what order, up to what amount, and what happens to the residue at maturity. The residue clause is the one that gets it signed — it says the unused reserve returns to the provider, and it converts the ask from a donation into a placement.

Layer two — the patient tranche. Subordinated, longer-dated, priced between one and four percent. This is where mission-aligned family offices, religious institutions, credit unions and Regulation A note programmes live. They want three things and they will tell you so: capital preservation first, a defined term second, a modest current yield third. In that order. The ask is not for generosity, it is for subordination and duration, and the price is the concession.

Layer three — the senior money. A bank, a CDFI, a state revolving fund. They are underwriting the structure now rather than the project: with ten percent of first loss and twenty percent of subordination beneath them, their attachment point is thirty percent. Say that sentence in the first meeting. A credit officer who hears attachment point hears someone who has done this before.

Layer four, optional and often decisive — the wrap. A public guarantee across the senior tranche. It is priced in basis points, it is applied for on a form, and it moves the internal rating of the loan by more than the fee costs.

The sequence, in weeks.

Weeks	Move	What you leave with
1–2	Size the book and compute the expected loss from a real history	The loss assumption, with its source
3–5	First-loss conversation, reserve not grant, residue clause drafted	A term sheet at the back
6–8	Patient tranche — subordination and duration, price last	A second term sheet
9–11	Senior lender, presented as a structure with an attachment point	Credit committee date
12	Guarantee application filed in parallel, not in sequence	A fee quote

Two governance rules hold the whole thing together. One waterfall document, and nobody's side letter contradicts it — a stack with three inconsistent descriptions of who gets paid first is not a stack, it is a future litigation. And the loss history is a single named source that every layer agreed to before pricing, because the argument you are about to have in year three will be about whether the losses were normal, and the time to settle what normal means is now.

DESTINY

How it holds when the person who assembled it leaves

A stack survives on three things, and all three are documentary rather than personal.

It revolves. Structure repayment back into the same facility rather than out to the balance sheet. At four percent lending, under one percent losses and about one percent administration, a corpus grows while it lends — one million dollars becomes about $6.9 million of cumulative lending across five seven-year cycles and the corpus is larger at the end than at the start. A facility that replenishes does not need to be re-approved, and a thing that does not need to be re-approved is a thing that survives a change of management.

The reserve provider gets a report they can use. One page, annually: capital mobilised, losses absorbed, reserve remaining, projects delivered. They are defending this internally too. Make the defence easy and the reserve renews; leave them to construct it themselves and it quietly does not.

The terms are boring. Novel structures die with their authors. A waterfall that looks like every other waterfall, a guarantee applied for on the standard form, an amortisation schedule a successor recognises at a glance — these are what allow the next person to administer it without understanding why it was built.

And here is where it fails. It fails when the first-loss layer is sized to convention rather than to a loss history, and the first bad year eats it. It fails when the patient tranche has a covenant the founder accepted verbally and the successor discovers in writing. It fails when everything is priced off a single senior lender's appetite, so one credit officer's departure re-opens the entire structure. And it fails — most often, and most avoidably — when the facility was assembled for one project rather than as a facility, so that at the end of the project the machinery is dismantled along with the scaffolding, and the next person starts from nothing.

DELIGHT

What it feels like

There is a specific pleasure in the meeting where the first-loss arithmetic lands. You say that most of the reserve is expected to come back, the programme officer's face changes, and they reach for a pen — not because you moved them, but because you handed them a number they can take to their own investment committee, who have been asking for exactly this and had no way to ask for it.

And there is the second pleasure, which is quieter and lasts longer: the moment the stack is assembled and you realise it is a machine, not an event. It will do the next deal without you. Somebody will use it in three years for something you would not have thought of, and they will not know your name, and the terms will still be right.

The work itself is good work. Reading a term sheet closely is one of the few forms of attention where every hour returns something — a clause that would have cost you, a definition that was doing different work than it appeared to. You will find yourself enjoying the waterfall drafting more than the pitch. That is the correct instinct. The pitch is what moves one person; the waterfall is what moves money for twenty years.

OPERATIONALIZE THIS

At the level of finance

The instrument: a blended revolving facility with a first-loss reserve and an optional guarantee wrap.

You are not raising a fund. You are assembling a facility — one legal entity or one account inside an existing one — that lends, is repaid, and lends again.

The structure.

Size. Twelve to twenty times your completed pilot, and no larger. The discipline: it must be deployable in twenty-four months by the people who already work here, without a hire you have not agreed.
Layers. Ten percent first loss at zero cost, twenty percent subordinated at one to three percent, seventy percent senior at market. Adjust the first-loss layer to three to five times expected cumulative loss on a named, sourced loss history, never to a round number.
Waterfall. Interest to senior, then subordinated; principal likewise; losses upward from the first-loss reserve. One document. Residue clause: unused reserve returns to its provider at maturity — this is what makes the reserve an investment rather than a gift, and it is what gets it approved.
Revolving mechanics. Repayments return to the facility for a defined availability period, typically five to seven years, before amortisation begins.
The wrap, where eligible. USDA OneRD, SBA 7(a), a state guarantee fund, or a multilateral guarantee for cross-border exposure. Applications run in parallel with the raise; they take longer than you think and they do not block.

Balance-sheet treatment. The first-loss reserve is an asset of its provider with an impairment allowance equal to expected loss — not an expense at placement. That single accounting distinction is why a foundation can post one dollar, expense roughly two cents of it a year, and keep the rest on the books. For the borrowing entity, the senior and subordinated tranches are debt; the guarantee fee is a finance cost amortised over the term, not a period expense. Where the facility funds an asset with a rising productive capacity, depreciate over its regenerated life, as in Chapter I.01. Bring your auditors in during the raise, not after.

The counterparty, in order. The community foundation or corporate foundation nearest you, for the reserve. Then two or three mission-aligned holders of patient capital — Regulation A note issuers, credit unions, religious endowments, family offices. Then your existing bank, and only then a CDFI or a state revolving fund if the bank is slow. Never open with the senior lender.

The first ninety days.

Day	Action	Artifact
1–15	Compute expected loss from a real, named history	The loss assumption, sourced
16–30	Size the facility and draft the waterfall, residue clause included	Draft term sheet
31–45	First-loss conversation, with the expected-cost arithmetic	Reserve term sheet
46–60	Patient tranche — subordination and duration, price last	Second term sheet
61–75	Senior lender, presented with its attachment point	Credit committee date
76–90	Guarantee application filed; blended cost recomputed on real quotes	The one-page stack

The number that decides it. One line, on the front page:

   book yield  −  expected losses  −  operating cost   >   blended cost of funds

Compute it on real quoted rates, not indicative ones, and compute it again with the senior tranche 150 basis points higher. If it does not survive that second computation, the first-loss layer is too thin — and the fix is at the back of the stack, never at the front. Raising the lending rate to close a 73 basis point gap does not save the structure; it selects the borrowers who will produce the losses that break it.

APPRECIATIVE QUESTIONS

Twelve, for a room

Discovery — what is already working

When has this organisation been lent money on terms that were genuinely patient — longer, cheaper or more subordinated than the market — and what made that lender willing?
Who in our network already sits behind somebody else's risk: a guarantee, a deposit, a letter of credit, a parent company comfort letter? What did it cost them, honestly, over the years they held it?
Which of our funders has money that is currently doing nothing between grant cycles — and what would it take for that to be sitting behind a lender instead?

Dream — what becomes possible

If our finance function could assemble a four-layer stack as routinely as it renews an overdraft, what would we fund next year that we are not funding now?
Imagine the reserve provider's annual page three years from now: capital mobilised, losses absorbed, reserve remaining. What numbers would we be proud to have on it?
If every asset here were financed on a tenor that matched how long it actually takes to regenerate, which decision would change first?

Design — what we build

What is the real loss history we would put in front of a first-loss provider, and who would have to agree that it is the right history?
Who is the one institution nearest us that could afford to be last in the waterfall — and what have they said yes to before?
What would we have to write into the residue clause for a reserve to feel like a placement rather than a gift to the person signing it?

Destiny — how it holds

What would have to be true for this facility to make its fourth loan after everyone who assembled it has moved on?
If our senior lender's credit officer left tomorrow, which parts of this structure would have to be renegotiated — and how do we reduce that list to none?
What is the first sign we would see that the first-loss layer was sized to a convention rather than to a loss rate, and who would notice it first?

WORKS CITED

Aleszczyk, A., Loumioti, M. and Serafeim, G. (2022). The Issuance and Design of Sustainability-linked Loans. Harvard Business School Working Paper 23-027.

Auzepy, A., Bannier, C. E. and Martin, F. (2023). "Are sustainability-linked loans designed to effectively incentivize corporate sustainability? A framework for review." Financial Management, 52(4).

Baker McKenzie (2021). In the Know: ESG Margin Ratchet.

Congressional Research Service (2023). Clean Water State Revolving Fund Allotment Formula: Background and Options. Report R47474.

Convergence (2025). State of Blended Finance. Convergence Blended Finance, Toronto.

Ding, L., Quercia, R. G. and Ratcliffe, J. (2008–2014). Community Advantage Program research series. UNC Center for Community Capital, Chapel Hill.

Enel S.p.A. (2019). Enel launches the world's first general purpose SDG-linked bond, successfully placing a 1.5 billion U.S. dollar bond on the U.S. market. Press release, 10 September 2019.

Environmental Finance (2025). Sustainability-linked loan decline accelerates.

Inter-American Development Bank and U.S. International Development Finance Corporation (2023). Ecuador Galápagos debt conversion for marine conservation. Transaction documentation and press materials, May 2023.

Iroquois Valley Farmland REIT, PBC (2023). Private Placement Memorandum, Offering of Series IV Soil Restoration Notes, and Regulation A annual reports (Forms 1-K and 1-SA), filed with the U.S. Securities and Exchange Commission.

Loan Market Association (2026). Horizons ESG: Market Outlook, Q1.

Opportunity Finance Network. CDFI industry portfolio performance: net charge-off rate. OFN member loan fund data.

Ostrom, E. (1990). Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge University Press.

Pew Bertarelli Ocean Legacy (2024). Innovative Financing Tool Helps Protect Galápagos Islands. Issue brief, December 2024.

The Chronicle of Philanthropy (1998). $51.8-Million Ford Grant to Assist Homebuyers.

U.S. Department of Agriculture, Rural Development. OneRD Guaranteed Loan Program: Business and Industry. Programme terms and fee notices.

U.S. Department of the Treasury, CDFI Fund. CDFI Bond Guarantee Program. Programme fact sheets and FY2025 programme overview.

U.S. Environmental Protection Agency. Clean Water State Revolving Fund: programme implementation and national information management system reporting.

Note on figures. Every figure in this chapter is computed in lib/verify/I_06.py and printed by python3 lib/verify.py I.06. Annual sustainability-linked-loan compliance cost is a stated assumption, priced at three levels with the sensitivity printed beside it; it is not a published figure and is not presented as one. The 0.58 percent net charge-off rate is the community development finance sector's own, and applying it to a book of a different composition is the first place this arithmetic would go wrong.