The one-way glass.

Which organisations reside in the chains of gender-marked money? We asked the owner of the largest chain.

Aggregate statistics stop one level above the chain. The OECD's databases can say how much gender-marked aid exists, but they carry no dimension that names who the money moves through, which we documented earlier in this series. So this article does the only honest next thing: it goes to the entity that owns the largest chain and reads what that entity publishes about itself.

Two publications. First, the transparency portal at open.unwomen.org, which is a good portal: budgets, expenses, donor tables, per-country pages, downloadable data. Second, the ledger UN Women publishes to the International Aid Transparency Initiative, IATI, a public standard where hundreds of organisations file machine-readable records of activities and transactions. The country-programme file runs to 58 megabytes and 1,167 activities. We recorded its cryptographic fingerprint and generation date in the source notes, so any reader can verify they are reading the same file we read.

The rule this initiative works under applies here with full force: every branch had to be a publishable story before we started. If the records named the receiving side, we would map it. If they did not, the absence would be the finding. There was no branch where the answer was nothing, which is the standard any question on this site has to meet before we spend a day on it.

The in-side of the glass is genuinely transparent, and the biggest name on it is the UN itself.

The portal's donor table for 2023 lists 562.87 million dollars in total contributions, which corroborates the audited figures this site already carries. The top of the table, in order: the UN Multi-Partner Trust Fund Office at 51.9 million, Sweden at 51.2, Germany at 39.8, Finland at 32.0, Norway at 30.9. The United States, the largest economy in the membership, sits eleventh at 18.8 million. In the portal's 2025 table the pattern sharpens: the Multi-Partner Trust Fund Office leads again at 70.8 million, ahead of every state.

Read that top line again. The largest funder of the UN's gender-equality entity is the UN's own pooled-fund machinery: money that member states place in common funds, which the funds' office then routes onward. The chain is real, it is intra-institutional, and it works. Nothing about it is improper. It is what assembled machinery looks like from the inside: an entity exists, so pooled funds have an address to route to. There is no equivalent address on the men's side, so no pooled fund routes anything there, and that absence is a fact about addresses, not about anyone's intentions.

One honesty note that matters more than the list: those are counts of activities a company is named on, not dollar amounts, because the ledger attaches funders to activities without assigning each funder a share. Anyone quoting this article with dollar figures per company is quoting something we did not say and cannot know from the public record.

And the standing rule of this site, stated where it belongs: none of this is a finding against anyone on the list. A company funding the women's entity is a company funding real work. The list is here because the question was who resides in the chain, and on the in-side, the entity's own records answer it completely.

Where the records show the work, the work is mostly what machinery is for.

It would be easy for an article like this to stop at tables, so here is what one chain looks like where the portal chooses to tell it. In Bangladesh, under a UN Women statistics programme and in partnership with the Asian Development Bank, the national statistics bureau built the country's first Household Production Satellite Account: an official measurement of unpaid domestic and care work. By the portal's own account, the work measured was worth 6.7 trillion taka in 2021, close to one fifth of the economy, 85 percent of it performed by women, and the finding reached the country's speech to the UN General Assembly.

That is what this site means by machinery. A thing nobody counted became a national statistical instrument with a budget line and a follow-up commitment, because an entity existed whose job it was to make the count happen. We cite it without reservation, and the sentence this initiative attaches to every such example holds: what is asked for is a second chain like this one, never a shorter first.

Notice one more thing about that story, because it matters in the next section. When the portal tells stories, it names the partners freely: the development bank, the statistics bureau, the ministries. The names exist. The telling of them is a choice the narrative pages make and the money pages do not.

On the out-side of the glass, the ledger names exactly one receiver, and it is the sender.

Here is the shape of the entity's public ledger, in full. 1,167 activities, and on every single one, the implementing organisation named is UN Women itself. 57,430 transactions from 2014 to 2023: 29,524 incoming-funds transactions worth 2.34 billion dollars, and 27,906 expenditure transactions worth 1.87 billion. Every transaction of both kinds names one receiver organisation: UN Women.

So the public money map is one-way glass. Money in carries 223 names, down to individual cosmetics brands. Money out, 1.87 billion dollars of it, carries one name, and it is the sender's own. The portal's data pages do not fill the gap: expense flows are published by category, region and theme, never by named counterparty. The narrative pages name partners in prose, as the Bangladesh story shows, but prose carries no amounts.

The claim ends exactly where the evidence ends, and we state its boundary as precisely as the finding itself: the receiving names are not in the portal's data pages and not in the entity's IATI ledger. That is the whole claim. Partner agreements may be disclosed in procurement records or board papers we have not found, and if a reader locates a public name-and-amount ledger of UN Women's receiving side, this article gets corrected, the correction gets dated and credited, and we will mean the thank-you.

We tried to make the wall an accident of one entity. It is bigger than that, which makes it worse as a fact and cleaner as a finding.

Before writing this we pulled a comparison file from UNDP, one of the largest publishers in the whole IATI system, and read its Afghanistan country ledger. The pattern is a variant of the same wall: its 1,771 expenditure transactions carry no receiver organisation at all, and only a handful of its commitments name downstream ministries. So the finding is not that UN Women behaves unusually. The finding is that the practice is systemic, and that UN Women is the without-exception case in the files we read: every transaction filled in, every receiver field pointing home.

The steelman also concedes the rulebook: the IATI standard makes the receiver field optional. No rule is being broken. And that concession is not a defence, it is the finding in its final form, because this site has seen this exact shape before. The SDG indicator on sexual violence in childhood covers both sexes on paper and has been measured for roughly 68 countries for women and roughly 12 for men. The indicator excluded nobody; the measuring did. The receiver field excludes nobody; the filling-in does. An optional field with no owner stays empty, in a survey and in a ledger alike.

That is the mechanism, and it is the article's actual subject. What gets counted becomes a problem with a name, a budget and a profession attached. The in-side of this glass is fully counted because every funder owns its own line and checks it. The out-side has no owner outside the entity itself, and ownerless fields are where this whole site does its work: the missing marker, the unmeasured indicator, the empty topic page, and now the unfilled receiver field are four instances of one phenomenon.

Which is why the ask was never a marker, and never a field. A marker counts money and a field holds a name, but neither reads a study, notices an absence, or answers for a chain. What is asked for is the apparatus that owns such things: an entity whose job it is to know, with the marker and the field among its cheapest bolts. The catalogue article showed every station of it already running somewhere. This article shows what the absence of an owner does to a ledger.

Money in has 223 names. Money out has one, and it is the sender's own.