@TITLE: Implicit Taxes and Measured Progressivity

@AUTHORS: Claude

@NOTICE: Experimental draft — not for citation or circulation · August 7, 2026

@AUTHORNOTE: Claude is an AI system developed by Anthropic; this paper was written by the Claude Fable 5 model generation. This paper was commissioned by Jason Furman.

@ABSTRACT: Municipal bondholders bear a tax no distribution table records: the yield they forgo because the exemption is capitalized into prices. Every maintained United States distributional framework counts the interest as income and records zero tax against it. We measure the implicit tax at the household level, combining Survey of Consumer Finances portfolios, marginal tax rates computed on the interest margin with TAXSIM, and a market-implied tax rate identified from the short end of the municipal term structure (38.0 percent, following Longstaff 2011). Correcting average effective federal income tax rates by percentile raises the 99th-to-99.9th percentile's rate by {{N:top1_delta_pp_headline:.2f}} percentage points at the anchor — {{N:grid_min_top1_delta_pp:.2f}} to {{N:wp_row_top1_delta_pp:.2f}} across the paper's grid of market-implied rates — with essentially no change below the 90th percentile, and adds ${{N:agg_implicit_tax_2021_soi_scaled_bn:.0f}} to ${{N:agg_implicit_tax_2021_bn:.0f}} billion of unrecorded burden for tax year 2021. The survey's aggregate tax-exempt interest exceeds the administrative record by a factor of {{N:scf_soi_ratio}}, and the correction is close to linear in that aggregate, so the same percentile correction computed on the IRS base is {{N:top1_delta_pp_soi_scaled:.2f}} points rather than {{N:top1_delta_pp_headline:.2f}}; the distributional shape is unchanged. Decomposing each holder's statutory rate into market-priced burden and inframarginal rent, we find the short-end market prices the average municipal bond dollar closely: aggregate rent in 2021 was within a billion dollars of zero under the two block-average constructions that pair holder and market rates on the same tax basis, and ran from −${{N:rent_fedcap_pairing_bn_abs:.1f}} to +${{N:rent_blockavg_hightax_bn:.1f}} billion across the full set of consistent pairings — small in every case against the burden. Rent is therefore not what the short-end anchor finds; it appears only under the lower rates that sit further out the curve, which this paper carries as stated scenarios rather than as readings of what 2021 holders were paid by maturity. Two neighboring corrections are sized here but not estimated: an owner-occupied housing correction would run the other way, landing in the middle of the distribution rather than at the top, and a corporate-equity correction cannot currently be signed at all. Neither is a bound, and neither is computed by percentile.

@NOTICE: Every number in this paper regenerates from public data by a single command; see the reproducibility statement in Section 4 and Appendix A5.

@PAGEBREAK
