Key takeaways
- BlackRock's lending agent takes 37.5% of the revenue in the Irish iShares range. In BlackRock's US ETFs, the fund keeps 82%.
- Vanguard's Irish umbrella paid its lending agent $1,134,721 against $13,518,041 of net lending income in the year to June 2025 — a 7.7% cut.
- Two S&P 500 trackers: one had 5.24% of net asset value on loan at its February 2026 year end, the other 0.016% at June 2025.
- Collateral at the iShares Core S&P 500 UCITS ETF covered 107.5% of the securities on loan at that date, and it was equities rather than cash.
- Lending added roughly 0.5 basis points to that fund, against an ongoing charge of 0.07%.
Your tracker lends its shares out, and it doesn't keep the whole fee
You've seen the line in a fund document — the fund may lend its securities — and you want to know two things. What happens if the borrower doesn't hand them back, and who gets the money?
Here's the short answer to both. The borrower posts collateral worth more than the loan, revalued every day, and the fund can sell that collateral if the borrower fails. The fee is then split between the fund and the manager. That split varies enormously by provider, and by fund within a provider.
In the Irish iShares range, the lending agent "will receive a fee of 37.5% of such securities lending revenue", according to the iShares plc annual report for the financial year to February 2026. In BlackRock's US ETFs, the same group's SEC filing says the fund "retains 82% of securities lending income". Vanguard's Irish umbrella paid its agent $1,134,721 against $13,518,041 of net lending income in the year to June 2025 — a 7.7% cut of the gross.
You won't find any of that on a factsheet. It sits in the annual report, because that's where the rules put it. ESMA's guidelines on ETFs and other UCITS issues say a fund's annual report should contain "the revenues arising from efficient portfolio management techniques for the entire reporting period together with the direct and indirect operational costs and fees incurred". The EU's securities financing transparency regulation goes further, requiring return and cost data "broken down between the collective investment undertaking, the manager of the collective investment undertaking and third parties (e.g. agent lender)".
A securities loan moves title against collateral, and the fund can call it back
The mechanics are plainer than the name suggests. The fund hands a block of shares to a borrower, usually a bank or broker covering a short sale or a hedge. The borrower pays a fee and posts collateral. Legal title passes, so the fund gives up the vote while the loan is open. Dividends are paid back to the fund under the contract.
Two features keep this inside the UCITS rules. The first is recall. DWS says every security on loan in an Xtrackers ETF "can be recalled daily / on demand", and that DWS "can recall a security to retain a proxy vote to participate in important or controversial votes". The second is collateral independence: ESMA requires collateral "issued by an entity that is independent from the counterparty".
Recall isn't a technicality. An ETF meets large redemptions by handing baskets of shares to authorised participants, so stock it can't retrieve is a problem. That plumbing is set out in our guide to how creation and redemption keeps an ETF's price near its NAV.
The manager's cut runs from 7.7% to 37.5% of the revenue
Here's what four providers disclose, ranked by how much of the gross revenue the fund keeps. The chart plots the same six figures.
| Fund range | Fund's share | Where it is disclosed |
|---|---|---|
| Vanguard Funds plc (Ireland) | 92.3% | Annual report, year to June 2025 |
| Xtrackers DAX ETFs | 91% | Securities lending policy |
| Xtrackers equity ETFs | 82% | Securities lending policy |
| iShares US ETFs | 82% | SEC filing, year to April 2026 |
| Xtrackers II fixed income | 70% | Securities lending policy |
| iShares plc (Ireland) | 62.5% | Annual report, year to February 2026 |
Read the top and bottom rows against the middle. The same manager, BlackRock, keeps 37.5% of lending revenue in its Irish range and 18% in its US range.
The two aren't measured identically, and the filing says so. The US 82% is a share of "securities lending income (which excludes collateral investment fees)". A separate charge for managing reinvested cash collateral sits outside it. The filing sets an all-in floor: the amount retained "can never be less than 70% of the total of securities lending income plus the collateral investment fees". Above a revenue threshold across the iShares ETF complex, the retained share rises to 85% for the rest of that calendar year.
Xtrackers publishes its rates directly. Its policy document says each fund receives "82% of the associated gross revenue generated (70% in case of Xtrackers II)", with the remaining "18% (30% in case of Xtrackers II)" split between the lending agent and the oversight entity. That schedule has applied since February 2024. Two DAX funds get 91%. So the rate isn't one house number even within a single provider.
Vanguard doesn't publish a rate at all. It publishes the arithmetic. Its Irish annual report tabulates net lending income and agent fees fund by fund, and the company totals for the year to June 2025 were $13,518,041 and $1,134,721. That's 7.7% for the agent and 92.3% for the funds — a figure you have to work out yourself.
The sums aren't trivial in aggregate. iShares plc booked £59.5m of securities lending income in the year to February 2026, already net of the agent's fee. Gross, that's about £95.3m, of which roughly £35.7m went to the agent.
Whether a fund lends at all matters more than how the fee is split
Take two funds tracking the same index. At its February 2026 year end, the iShares Core S&P 500 UCITS ETF had 5.24% of net asset value out on loan. At its June 2025 year end, the Vanguard S&P 500 UCITS ETF had $11.1m on loan against net assets of $71.3bn — 0.016%.
The income follows. The iShares fund earned $1.04m of lending income for the year, net of the agent's 37.5%, on net assets of $20.0bn. The Vanguard fund earned $43,864. One fund applied a generous split to almost nothing. The other applied a harsher split to a real programme.
Keep the magnitude in view all the same. That $1.04m is roughly 0.5 basis points of the iShares fund's net assets, against an ongoing charge of 0.07%. Lending revenue is one of the small credits that lets a physical tracker close the gap to its index, and the full accounting of that gap is in our guide to tracking difference versus tracking error.
The collateral covers the borrower failing, and it's sized above the loan
Cover ran above the loan in both funds. At the iShares fund's February 2026 year end, collateral received equalled 107.5% of the securities on loan. At Vanguard's June 2025 year end, cover on the S&P 500 fund was 106.6%.
In the US the minimum is written into the fund's own policy. Initial collateral has to be "at least 102% of the current market value of the loaned securities for securities traded on U.S. exchanges and a value of at least 105% for all other securities", then "maintained thereafter at a value equal to at least 100% of the current market value of the securities on loan". Xtrackers applies the same floor in Europe, at "at least 100% of the global valuation of the securities lent", marked to market daily.
There's a second layer. The iShares plc report says each fund "benefits from a borrower default indemnity provided by BlackRock, Inc.", that the indemnity "allows for full replacement of securities lent", and that BlackRock "bears the cost of indemnification against borrower default".
What the collateral doesn't cover
Five gaps, all of them disclosed.
First, the price of the collateral between default and sale. DWS spells this out. If a borrower fails to return securities, "there is a risk that the collateral received may be realised at a value lower than the value of the securities lent", for reasons including "inaccurate pricing of the collateral, adverse market movements in the value of the collateral, intra-day increase in the value of the securities lent".
Second, the indemnity's edges. DWS says its agent's obligation to indemnify "is limited to the event of an act of insolvency in respect of a borrower". In a default outside that definition, with a simultaneous collateral shortfall, "the Fund will suffer a loss".
Third, the law where the borrower fails. BlackRock's US filing notes that "bankruptcy or insolvency laws of a particular jurisdiction may impose restrictions on or prohibitions against such a right of offset" — the right of offset the whole structure depends on.
Fourth, what the collateral is made of. The iShares plc report says the collateral its funds received "consists of shares admitted to dealing on a regulated market". Vanguard's report says collateral "is limited to high quality sovereign debt (U.S. Treasuries, U.K. Gilts, etc.)". Equity collateral against equity loans is likelier to fall on the day you need to sell it. ESMA requires collateral independent of the counterparty, not independent of the market.
Fifth, cash collateral reinvestment, which European funds have largely closed off. The FSB flagged it as a source of maturity and liquidity transformation that "can present risks and negative externalities to firms beyond the beneficial owner or agent lender in a stress event", and sized the global activity at "$1.0 trillion in Q3 2008". DWS says its funds "will not engage in any reinvestment of collateral". BlackRock's US funds do reinvest, in money market funds run by an affiliate, which is why their revenue split is defined to exclude collateral investment fees.
Scale is the thing collateral doesn't change at all. At its February 2026 year end the iShares $ Treasury Bond 1-3yr UCITS ETF had 96.15% of net asset value out on loan, and 102.42% of what the report counts as lendable assets. Xtrackers caps direct-replication funds at "up to 50% (or up to 100% for Xtrackers II Funds) of its Net Asset Value at any one time". A collateralised loan is still a loan.
A swap-based tracker avoids this entirely, because it doesn't hold the index constituents to lend. It exchanges the exposure with a bank instead, which relocates the counterparty risk rather than removing it — that trade is set out in our guide to physical versus synthetic replication.
The best case for the manager's share
The obvious objection to a 37.5% cut is that the fund owns the assets and the agent merely arranges the loan. The disclosures answer that partly.
The agent absorbs the running costs. iShares plc says the agent "will pay any third party operational and administrative costs associated with, and incurred in respect of, such activity, out of its fee", and that where those costs exceed the fee, the agent "will discharge any excess amounts out of its own assets". Vanguard says the same of its agent: "All operational costs in support of the lending programme are borne by the lending agent." BlackRock also carries the cost of the borrower default indemnity, which is a balance-sheet commitment rather than a service.
The counter to that is Xtrackers. Its DAX funds return 91% of gross revenue and its equity funds 82%, from the same agent under the same document. Whatever it costs to run a lending programme, that cost doesn't explain the 28.5-point gap between 91% and 62.5%.
What these numbers don't tell you
The on-loan percentages are snapshots at one date, not averages for the year. A fund can sit near zero for eleven months and look busy at year end, or the reverse. The revenue figures cover a full year, so pairing them with a year-end net asset value gives a rough rate rather than a precise one.
The splits aren't measured on one basis either. The Irish iShares 37.5% is a share of lending revenue. The US 82% excludes collateral investment fees, and the comparable all-in floor there is 70%. Reading the two as a clean comparison overstates the difference between them.
Lending revenue is lumpy, because most of it comes from a small number of hard-to-borrow securities. The iShares MSCI Korea UCITS ETF earned $661,000 of lending income in the year to February 2026, against $118,000 the year before. One year is a small sample, and a fund's revenue can move by a multiple with nothing about the fund having changed.
And the collateral machinery hasn't been tested by a large borrower failure since these rules were written. What the disclosures describe is a design, not a track record.
What would change the picture
Three things, in order of how quickly you'd see them.
The split is a contract, and contracts change. Xtrackers dates its current schedule to February 2024, so the rate in this year's report needn't be the rate in next year's. The line to check is the related party note, not the factsheet.
A borrower default that ate through collateral would settle an argument that is currently theoretical. Until one happens, the case for the collateral rests on daily marking, a buffer of 6.6 to 7.5 points at the two funds above, and an indemnity with a defined edge.
And the amount at stake moves with lending intensity, not with the split. A fund with 0.016% of assets on loan is barely in this business. A fund with 96.15% is in it to the neck. That number is published once a year, in the same document as the split, and it's the one worth reading first.