The Tax-Free Thing You Own

Municipal bond funds: eighteen products in one shelf slot, and what is actually inside the credit.

Eighteen national intermediate-term tax-exempt bond funds from eighteen sponsors, holding $236,462,667,017.86 of net assets between them, read from their own regulatory filings and, where one exists, from the fund's own annual report. No aggregator, no rating file, no fact sheet, no vendor data. Filing periods run 2026-04-30 to 2026-06-30 and are printed beside every figure in the full report.

This is about what the funds hold. It says nothing about any reader's tax position.

The finding

Funds sold into the same slot hold visibly different credit. Health care and senior living runs from 0.37% of classified municipal value in one fund to 22.27% in another — a factor of sixty-one. Housing runs from 0.06% to 19.41%. Credit lent on through a public shell to a named private borrower runs from 17.45% to 50.69%: one dollar in six in one fund, one dollar in two in another.

A "national" fund is national to very different degrees. The largest single state runs from 9.35% to 30.10% of classified value, and the top three states from 20.74% to 53.26%.

And the index funds are the concentrated ones. The three index funds in the set are the three most concentrated by top-three sector — 74.93%, 73.20% and 72.94% — and the three most New-York-heavy. The active funds spread wider and disagree with each other about where to go.

What we did not expect, and it is the better finding

Whether a fund's own filing can tell you what you own is itself a variable, and it varies more than the portfolios do.

Sector can be read from 74.23% to 98.31% of a fund's municipal value; state from 56.85% to 99.23%. But the split every municipal buyer is taught to ask about — a general obligation backed by taxing power against a revenue bond backed by one project's receipts — can be read from 99.51% of one fund's value and 3.25% of another's.

For fourteen of the eighteen, that split cannot be computed at all. Not because a fund is hiding anything: because of how its sponsor writes a security description into a form. On 96.75% of one fund's municipal value, the filed text simply does not contain the answer.

That is a property of the filer's vocabulary, not of the credit. We print the classification rate beside every composition figure rather than behind it, and where the rate is too low we print no figure at all.

The three numbers

What the regulatory file does answer

It carries none of the five fields this research set out to measure — no sector, no state, no bond type, no rating, no tax status. It carries three that no fund-comparison page prints:

• What the holdings sum to against net assets. On this shelf, 98.028% to 101.786%. Every fund on it is unlevered.
• What the fund owes. 0.207% to 2.950% of net assets, all of it unsettled trades and payables.
• Which holdings the fund's own administrator has flagged as defaulted or in arrears. Eleven of the eighteen flag at least one default; seven flag interest in arrears. The largest defaulted share is 4.115% of net assets and the largest arrears share 1.841%.

The largest fund on the shelf carries seven lines of one obligor — $31,180,000 of face carried at $3,118.00, one cent on the hundred dollars, three of the seven already past their stated maturity date. Its annual report names the project. Its monthly filing names only the financing authority.

A high-yield municipal fund is a different animal, and this report refuses to average it in. The one used as a labelled contrast has holdings summing to 129.145% of net assets, liabilities at 34.998%, a third of the fund in restricted securities, 9.174% of net assets flagged defaulted and 7.252% in arrears. No figure about it is comparable to any figure about the eighteen.

What we cannot tell you

• What any of these funds will do next. There is no forecast anywhere in this report.
• How two of them overlap. Security-level overlap between funds was not measured.
• What a fund's credit rating profile is. No cross-fund ratings column exists from filings: one issuer publishes a credit-quality table as an image and another publishes none across a 124 MB document. We print no rating.
• Anything about your tax position. Nothing here is a yield, a fee, a return, an after-tax outcome, or tax advice.

What this report corrected in its own work

Its own scope document said two filings pairs reported different net assets on the same date. There are three — 0.142%, 0.172% and 0.203% apart — and all three are printed with both sources named and neither preferred. A committed parser's "9 of 9" reconciliation score was 8 of 9 when re-run; the defect was found, fixed, and printed. A first cut of the state rule would have invented a Colorado concentration out of nothing, because "CO" abbreviates "County" on 1,413 of the 1,540 filed lines where it fires. Ten state codes were refused rather than guessed, and the ten states that understates are named.

1,741 automated checks pass. 64 planted defects were caught, all of them, with no survivors and no harness failures.

Disclosures. The author holds no position in any fund or security named here. No issuer payment, no advertising. Impersonal research, identical for every purchaser, tailored to no one. No statement about what any reader should hold, no recommended allocation, no forecast, and nothing about anyone's tax position. Finnacl Research is not an investment adviser.