The short version
- Our effective-holdings measure is computed from every position a fund reports on SEC Form N-PORT. For IWM that produced 33.2 — a 2,017-stock index fund reading as concentrated as 33 equal positions.
- The reason is one line in the filing: a $14.1 billion position in BlackRock Cash Funds, 17.0% of net assets, marked
isCashCollateral = Y. It is cash the fund holds as collateral for securities it has lent out, matched by a liability of the same size. It is not an investment the fund chose. - Computed on the fund's own investments, IWM behaves like about 840 holdings. The published 33.2 was a statement about the collateral, not the portfolio.
- Of the 694 funds whose position-level filings we held this morning, 461 carry some cash collateral and 79 carry 5% or more of net assets. For 73 funds the collateral sleeve produces a quarter or more of the concentration figure. As of today those 73 are marked not comparable on this site, with the reason and the ex-collateral figure beside the headline.
- The headline definition and the quarterly history have not been changed yet. That is the next job, and it is recorded below rather than hidden.
How we found it
The daily video pipeline picked PRFZ, the Invesco RAFI US 1500 Small-Mid ETF, and rendered "1,540 holdings, behaves like 16." A 1,500-stock fund behaving like sixteen is not a finding; it is a symptom. The largest position in PRFZ's filing is the Invesco Private Prime Fund at 23% of net assets, and the filing's own securities-lending block reads:
cashCollateralCondition isCashCollateral="Y" cashCollateralVal="663255310.90"
The same pattern sits at the top of IWM's filing:
| Position | % of net assets | Asset category | Filing flag |
|---|---|---|---|
| BlackRock Cash Funds: Institutional, SL Agency Shares | 17.02% | STIV (short-term investment vehicle) | isCashCollateral = Y |
| Moog, Inc., Class A | 0.38% | Equity | on loan: $10.6M |
| Hut 8 Corp. | 0.37% | Equity | — |
Two things follow. First, the long positions in IWM's filing sum to 116.8% of net assets, not 100%; the extra 16.8 points are the collateral, which the fund holds against loaned stock and owes back. Second, the largest "holding" is not a holding in the sense the concentration measure assumes.
Why one position breaks the count
Effective holdings is the inverse Herfindahl index: one divided by the sum of squared position weights. It is the number of equal-sized positions that would give the same concentration. The arithmetic is dominated by the largest weights.
A 17% position contributes 0.17² = 0.0289 to the sum of squares. IWM's two thousand stocks, the largest of them 0.38%, contribute about 0.0012 between them. The collateral line alone is 96% of the total. Divide one by the total and you get 33; drop the collateral line, renormalise the stocks to 100% of what remains, and you get about 840.
This is why the effect is largest in funds that lend the most relative to their size — small-cap and bond index funds — and why it barely touches VOO (no collateral position filed) or IVV (0.17%).
How wide it is
The fourteen largest funds on this site where collateral produces a quarter or more of the concentration figure, from the filings we hold:
| Fund | Assets | Filed holdings | Filed "behaves like" | Collateral, % of net assets | Own investments behave like | As of |
|---|---|---|---|---|---|---|
| IJH | $129.1B | 407 | 202.1 | 3.84% | 287.4 | 2026-06-30 |
| IJR | $117.5B | 656 | 173.5 | 5.45% | 356.1 | 2026-06-30 |
| VXF | $97.8B | 3,372 | 329.2 | 3.63% | 582.2 | 2026-06-30 |
| IWM | $97.1B | 2,017 | 33.2 | 17.02% | 840.2 | 2026-06-30 |
| RSP | $91.3B | 507 | 340.3 | 3.91% | 492.9 | 2026-04-30 |
| LQD | $34.2B | 3,137 | 173.9 | 7.23% | 1,889.9 | 2026-05-31 |
| IGIB | $22.3B | 2,990 | 78.2 | 11.07% | 1,843.2 | 2026-05-31 |
| USIG | $20.7B | 11,460 | 110.8 | 9.42% | 6,893.6 | 2026-05-31 |
| HYG | $20.2B | 1,323 | 58.3 | 12.69% | 907.5 | 2026-05-31 |
| IWO | $17.7B | 1,137 | 41.4 | 14.75% | 409.7 | 2026-06-30 |
| DFAS | $15.6B | 2,120 | 104.9 | 11.34% | 908.7 | 2026-04-30 |
| IWN | $15.3B | 1,426 | 223.8 | 5.05% | 519.9 | 2026-06-30 |
| SCZ | $14.9B | 2,052 | 412.8 | 3.82% | 1,027.2 | 2026-04-30 |
| DFAT | $14.5B | 1,279 | 165.9 | 5.87% | 385.5 | 2026-04-30 |
A small sleeve is enough when the rest of the portfolio is very flat: IJH's 3.84% moves the figure from 202 to 287, because an equal-ish 400-stock portfolio has almost no squared weight of its own for the collateral to compete with.
Funds below the quarter threshold keep their headline but now carry a note. VXUS reads 259.8 filed, 280.1 on its own investments (1.74% collateral); VWO 39.1 against 41.0; VEA 258.3 against 268.9; ITOT 65.0 against 66.0.
What changed on this site today
- Every fund record now carries three new fields read straight from the filing: the collateral share of net assets, the share of the concentration figure the collateral sleeve produces, and the effective holdings of the fund's own investments with collateral set aside.
- A fund whose collateral sleeve produces 25% or more of the figure is marked not comparable, on its page, in its search snippet, in the data files and through the MCP server, with this reason and the ex-collateral figure stated. Seventy-three funds are marked today.
- Funds with 1% or more collateral below that threshold carry the note on their page. One hundred and sixty-nine funds do.
- The methodology page states the rule, dated.
- A short video we had published on USIG, giving 110.8 as its figure, was taken down. Its own investments behave like about 6,894; the 110.8 was 98% collateral.
What has not changed yet, and why
Update, 2026-09-18. This is now done. Every historical filing was re-measured on the same definition (the originals are archived so this never needs repeating), and since 2026-09-18 the headline figure on every fund page, in the data files and through the MCP endpoint is computed on the fund's own investments with collateral set aside; the as-filed figure is kept beside it under asFiled. IWM's page now reads 840, not 33. The paragraph below is left as it was written on 2026-09-14.
The headline figure on every fund page is still computed from the filed positions as filed, collateral included, and so is the quarterly history back to 2019. We have not switched the headline to the ex-collateral definition today because the history is built by re-reading every past filing, and switching the present without the past would put a chart on each page whose last point disagrees with the number above it. The re-measurement runs next; when it lands, the definition changes everywhere on the same day and the methodology page will carry the date.
Until then, the rule of thumb for reading this site is simple: if the box above the figure says not comparable, read the number in the box, not the headline.
What this can and cannot tell you
- Securities lending is a normal, disclosed programme; the collateral is the fund's to hold and return. Nothing here is a claim about lending risk, income, or whether a fund should lend.
- Collateral levels move with demand to borrow the fund's stocks, so the share reported here is as of the filing date shown and will differ next quarter.
- The ex-collateral figure renormalises the remaining filed positions to 100%. Other cash-like positions that are not collateral stay in.
- The flag we rely on,
isCashCollateral, is what the fund itself reported. Where a fund reports collateral without the flag, we would not catch it; we have not found such a case, which is not the same as saying there is none.
Sources: IWM's filing, SEC EDGAR primary_doc.xml, period 2026-06-30, filed 2026-08-25; every figure above is in concentration.json under cashCollateralPct, collateralHhiShare and effectiveHoldingsExCollateral.