The Ledger Nobody Keeps

In 1996, the Inter-American Development Bank sent me to Bolivia to assess what it would take to strengthen the country's housing finance market, including what would be required to develop a secondary mortgage market. That work included evaluating the thirteen mutuales de ahorro y préstamo that were the country's primary originators of housing loans, and assessing which institution could sit above them as a conduit.

The mutuales were member-owned, funded by local deposits, rooted in the departments they lent in. Nationally they held about nine percent of the financial system's deposits. In Pando that figure was thirty-four percent, and in Beni twenty, and in those departments they were, for many households, the only formal savings institution within reach.

The secondary mortgage market never happened.

What did happen took twenty-eight years, and nobody was measuring for it.

I want to tell that story, because it makes an argument about development finance that I think is right, and because it makes it in a way that does not depend on today's aid crisis. Arguments that depend on crises expire when the crisis does.

The argument everyone is making right now

Official development assistance from OECD DAC members and associates fell 23.1 percent in real terms in 2025 on the OECD's headline grant-equivalent measure, the largest annual contraction on record. The annual financing gap for the Sustainable Development Goals sits somewhere around $4 trillion. No plausible recovery in donor budgets closes a gap that size. From this, a conclusion is being drawn across the sector: since the money is gone, we must build local institutions that can raise and deploy capital themselves. I agree with the conclusion. I distrust the reasoning, because it makes locally led finance sound like what you settle for when the grant ends.

In 1996 there was no aid crisis. The IDB was funding the work. Capital was not the binding constraint in Bolivia, and the program failed anyway, for reasons that had nothing to do with how much financing was available. The contraction of 2025 does not create the case for building local financial institutions. It removes the excuse for not having built them thirty years ago.

What imported capital does, and what it does not

Development finance institutions have become genuinely good at mobilizing private capital alongside public money. The volumes are real and they are growing.

The distribution is the problem. The large majority of mobilized private finance lands in middle-income countries, and only a small fraction reaches low-income ones. That is not a failure of intent. It is what happens when the ultimate credit decision sits thousands of miles from the borrower and the market in which that borrower operates.

Risk that cannot be assessed locally gets priced as unknowable, and unknowable risk gets avoided.

The instruments follow from this. Blended structures are best suited to moving relatively large tickets to counterparties that can become bankable with the right allocation of risk. That leaves a very different financing problem largely untouched. The IFC puts the financing gap for formal small and medium enterprises in developing countries at roughly $5.2 trillion, and finds that seventy percent of MSMEs in emerging markets lack the financing they need to grow. International DFIs are not structured to underwrite a $40,000 working-capital loan to an agro-processor in a secondary city. A local bank or microfinance institution can, and does, if it has the balance sheet, the liquidity and the risk systems to do it at scale.

There is a second problem, less visible in the aggregates. Programs built around external capital and external management often leave too little capability behind when the funding cycle ends. Systems remain with the donor or implementing partner; institutional knowledge leaves with the external team. The humanitarian sector has run this experiment at scale. A decade after the Grand Bargain committed its signatories to channel at least 25 percent of humanitarian funding to local and national actors as directly as possible, the figure remains below 4 percent.

The commitment was sincere. The plumbing was never built.

De-risking a transaction, and de-risking an institution

Blended finance is the tool most often proposed for this, and it is a genuinely useful one. Convergence recorded 123 deals reaching close in 2024, worth roughly $18 billion, against a cumulative market of about 1,350 transactions and $249 billion.

But blending concessional and commercial capital is a technique, not a theory of change. The same structure can be used two very different ways.

The first uses concessional capital to de-risk a transaction so an international investor will participate. The deal closes, the project gets built, the mobilization ratio goes in the annual report. Nothing about the local financial system is different the day after.

The second uses concessional capital to de-risk an institution. First-loss tranches that let a local bank extend tenors it could not otherwise price. Partial credit guarantees that allow a domestic lender to build a track record in a new sector and eventually underwrite it unsupported. Technical assistance aimed at the origination and risk-management capability that makes the next deal possible without a guarantee at all.

The test for any blended structure is simple. If the concessional layer were removed in five years, would the local institution be able to do more than it could before the transaction? If the honest answer is no, the structure bypassed the financial system rather than strengthening it.

That test is easy to state and harder to apply than it looks, because even when you know what capability you are trying to build, you may not know what it will ultimately be used for. Bolivia taught me that.

What happened in Bolivia

The thirteen mutuales were, in varying states of financial health, functioning lenders. Several carried capital deficiencies serious enough that the government forced liquidations and mergers within the year. The central institution above them, the Caja Central de Ahorro y Préstamo para la Vivienda, was liquidated by the banking superintendency on 22 January 1997.

The secondary market that all of this was meant to produce never formed. It is worth being precise about why, because the reasons are not the expected ones and every one of them survives to the present.

The capital constraint dissolved. By the late 1990s Bolivia's commercial banks were highly liquid. They did not need to sell mortgage portfolios to a conduit, because they had their own funding and preferred to keep the interest on their own books. The entire structure had been designed to relieve a shortage of long-term capital, and the shortage went away before the secondary mortgage market was fully established. This cuts against the way the case for local financial institutions is usually made, including the case I have just made — and it should. This is not fundamentally an argument about scarcity. Scarcity makes it urgent. It does not make it true.

The currency problem outlived its own solution. In 1995 the mutuales' mortgage books were overwhelmingly denominated in US dollars, lent to Bolivians who earned bolivianos. The older maintenance-of-value contracts had largely been phased out of new lending, and there was no alternative: the inflation-indexed housing unit that would have made local-currency mortgages workable did not exist until late 2001. The mismatch was not a failure of judgment by the lenders. The instrument was missing.

What is instructive is what happened after it arrived. The indexed unit was created, and the mortgage market stayed dollarized anyway, for years, starving the conduit of a uniform asset pool. Bolivia did eventually de-dollarize, and dramatically: from one of the most dollarized financial systems in the world in the early 2000s to one of the least by the middle of the following decade. It did not get there because a better instrument became available. It got there through a decade of unglamorous domestic prudential policy. Reserve requirements weighted against foreign-currency deposits. Capital and provisioning charges on foreign-currency lending. A tax on foreign-exchange transactions. Caps on dollar deposit rates.

That sequence should temper how we talk about currency risk. Facilities that let borrowers hedge into local currency, like the Currency Exchange Fund, are valuable and I have recommended them. But they make local-currency intermediation possible. Bolivia is the reminder that possible and actual are separated by roughly ten years of domestic regulatory work that no external facility can do on a country's behalf. There is a hierarchy here that the sector rarely acknowledges: you can move currency risk onto a balance sheet built to hold it, or you can take it out of the system. The second is better, and no one can do it for you.

The credit processes were not interoperable. This is the finding I would most want a development finance institution to take from the episode. Even the sound mutuales underwrote differently from one another, appraised property differently, and documented differently. There was no common standard. Pooling their mortgages meant pricing thirteen underwriting cultures at once, which made the pool too expensive to evaluate against any recognized secondary-market standard.

We found this during our reviews and said so. What was missing was the institutional machinery to act on it. An evaluation could identify the constraint, but neither the proposed conduit nor the program had the authority to require thirteen independent member-owned lenders to adopt common standards. That would have required a regulator willing to mandate them, or an apex institution with enough leverage to make them a condition of access. Neither was in place in the window of time when it would have mattered.

That is worth sitting with, because it is not the failure people expect. The diagnosis was correct but there was no mechanism to address it. Thirty years on, that is the constraint that actually bound.

And the choice of which institution to strengthen was not merely a technical choice. The incumbent central institution had the relevant experience, including with asset-backed structures. It also carried legacy liabilities and no political support. The vehicle that replaced it was built new, staffed with capital-markets specialists recruited for the purpose, and pointed at Bolivia's newly reformed pension funds as a source of institutional demand.

Building new was almost certainly the more executable option. It is worth naming what it cost. The accumulated experience was liquidated along with the balance sheet, and the capability had to be bought back from the market. Building a greenfield institution is frequently chosen because it is cleaner rather than because it is better, and "we strengthened local institutions" and "we built a new institution locally" are not the same claim, although they are reported the same way.

What was still there

By its own terms the program failed. The institution did not.

NAFIBO, the second-tier vehicle built to be the mortgage conduit, took the securitization framework it had been given and applied it everywhere else. It came to dominate Bolivia's securitization market. It pioneered securitizations of future cash flows for small firms and agricultural cooperatives, structures that were novel globally rather than only locally. It worked as an apex wholesale lender channeling international liquidity into Bolivia's regulated microfinance institutions, within the financial ecosystem that made Bolivia an international reference point for microfinance. In 2007 it was restructured into the Banco de Desarrollo Productivo, the national development bank.

In October 2024, the BDP was accredited to the Green Climate Fund as a direct access entity: able to receive international climate finance without a multilateral intermediary standing in between.

That is what localization commitments across this sector say they are trying to produce. It took twenty-eight years. It grew out of a housing finance program that failed at the thing it was designed to do, and no mortgage-focused results framework in 2002 would have recorded a single step of it, because the framework asked about mortgages.

Apply the five-year test to that and it passes, at something nobody specified. Which is the refinement I would make to my own rule: build adaptable capability rather than designing it too tightly around the transaction that happens to justify the investment today. The mandate is usually the least durable thing in the design.

The other side of the ledger

The thirteen mutuales I encountered in 1996 have dwindled, through failure, legal transformation and merger, to a successor category with three licensed institutions. ASFI's latest posted register lists three entidades financieras de vivienda: La Primera, La Promotora and El Progreso. Some of that was necessary. Mutual Guapay was a genuine failure: it was intervened in January 2008 after reporting negative equity, and its assets and liabilities entered the statutory resolution process. La Paz — by then an EFV — was intervened in May 2016. Those were the system working.

But the institutions in Pando, Beni and Potosí followed a different path. They survived the 1996 restructuring, the dollarization and the crisis years. In 2013 Bolivia passed a new financial services law that required the mutuales to transform into a new legal form, and in November 2015 the survivors became entidades financieras de vivienda, or EFVs. On December 3, 2018, ASFI revoked the licenses of the Pando, Potosí and Paitití EFVs as a consequence of their merger by absorption into La Promotora. They were dissolved without liquidation.

Nothing dramatic happened to them in the sense that matters here. They were not intervened for insolvency. They were absorbed into a larger institution, headquartered in Cochabamba, that continues to serve Pando, Beni and Potosí. What disappeared was not access to regulated finance. It was three stand-alone, locally headquartered housing-finance institutions.

That is the part I find hardest to fit into our usual ledger. The record does not let me say that regulation caused the consolidation. It does show that the consolidation occurred inside a regulatory modernization I would otherwise regard as necessary. The system could become more standardized and more prudentially coherent while becoming less locally rooted at the same time.

The system-level data record the surviving balance sheets. They do not record the loss of a locally headquartered institution as a development outcome.

What to measure instead

This is why I am wary of the percentage targets that dominate the localization debate, even though I think the people who set them were right to try.

USAID's attempt was the most serious one. The agency set a target of 25 percent of funding to local partners and published its own data against it. It reached $2.1 billion in FY2024, or 12.1 percent of its acquisitions and assistance funding, up from 10.5 percent the prior year and 11.3 percent the year before that. On its parallel goal of half of programs led by local actors, 35 percent of activities qualified. New awards to local and regional partners were up 87 percent since FY2021.

Read one way, that is a target missed, and a series that went backwards before it went forwards. Read another, it is the only donor in the system that defined a measurable commitment, tracked it publicly, published a number that moved the wrong way, and kept publishing.

The problem is not that the target was too ambitious. It is that a percentage measures the flow of funds and not the durability of what the funds build. An agency can hit 25 percent and leave local institutions no stronger than it found them, and it can miss badly while building something that lasts thirty years.

For investments intended to strengthen local financial institutions, four things I would rather see in a results framework:

Interoperability of credit processes. Not whether each counterparty underwrites soundly in isolation, but whether their processes are standardized enough that portfolios can be pooled, rated, sold or refinanced by someone else. This is the constraint that killed the Bolivian secondary market and it is almost never in a logframe.
Unguaranteed origination after the guarantee expires. Measured at twenty-four and thirty-six months in the target sector. This is the five-year test in its directly observable form.
Growth in the counterparty's own balance sheet and, where applicable, its deposit base, rather than disbursement against the facility.
Whether the credit process that was built is still in use, and staffed by people the institution pays.

What I would tell a fund manager

Resilient financial systems are not built by foreign capital alone. They are built through banks, funds, regulators and credit infrastructure capable of holding domestic savings and lending them productively. Foreign capital is more useful — and more likely to stay — when those institutions exist.

The current contraction in aid budgets is not a reason to defer that work. It is the reason to prioritize it. But it was worth doing when the money was plentiful too, and the fact that we mostly did not is the more damning observation.

Two ledgers came out of a failed housing finance program in Bolivia. On one, an institution that compounded for twenty-eight years into something that can now access international climate finance directly, which is more than anyone designed and more than anyone tracked. On the other, three departments that lost their stand-alone, locally headquartered housing-finance institutions when they were absorbed into a larger regional institution — not through an insolvency intervention, but through merger. Both are real. Only one of them gets counted.

Notes on sources

Figures on official development assistance are from the OECD's preliminary 2025 ODA data, released April 9, 2026. The 23.1 percent decline is the real-terms change in headline ODA from DAC members and associates on the OECD's grant-equivalent basis. Blended-finance volumes are from Convergence's State of Blended Finance. The MSME financing gap and the seventy percent figure are from the IFC. Grand Bargain localization figures are from the Inter-Agency Standing Committee and subsequent independent analysis. USAID localization figures are from the agency's own published data as analyzed by Publish What You Fund.

Bolivian regulatory events are from Law No. 393, ASFI records and ASFI-sourced official government statistics. Law No. 393 required the transformation of mutuales into EFVs; ASFI's statistical series dates the transition to November 23, 2015. Guapay's 2008 intervention is documented in ASFI/SBEF data and a contemporaneous Banco Central de Bolivia resolution. La Paz EFV's 2016 intervention is documented by ASFI Resolution 302/2016. The 2018 consolidation is documented by ASFI Resolutions 1552/2018, 1553/2018 and 1554/2018, which revoked the licenses of three EFVs as an effect of their merger by absorption into La Promotora. ASFI's latest posted register, updated April 30, 2026, lists three licensed EFVs. Bolivia's de-dollarization and the historical deposit shares by department are documented by the Banco Central de Bolivia. Recollections of the 1996 evaluation, the scope of the housing-finance assignment and the market studies that accompanied it are my own.