The Shape of the Tax: Statutory Incidence, Notches, and the Design of New York City's Mortgage Recording and Transfer Taxes

New York City charges a mortgage recording tax when a buyer borrows, while a cash purchase incurs no mortgage recording tax. Using 172,633 observed purchase loans, this paper measures the borrower-borne charge at 1.41% of the purchase price for houses and 1.28% for condominiums. It then asks how a payment-neutral replacement should be shaped. New York's 2019 transfer-tax reform introduced discrete jumps at several prices. Transactions shift toward prices just below measurable buyer-paid thresholds. Point estimates for the USD 37,500 jump at USD 5 million and the USD 5,000 jump at USD 2 million are similar, and a response proportional to the buyer's rate is rejected. The evidence does not establish equal responses or settle whether statutory incidence matters. The city's USD 500,000 threshold is excluded from avoidance inference because its lower rate also applies at exactly USD 500,000. The paper costs continuous schedules that would replace about USD 246 million a year in borrower-borne tax within its four-borough, USD 300,000 to USD 6 million sample. These are static accounting calculations, not an optimal-tax estimate or a forecast of prices, transaction volume, or actual revenue after reform. Condominium loan amounts in the replacement calculation are imputed from the measured average burden. Read the manuscript PDF first; the supplemental appendix holds robustness tables. paper_in_brief.md gives a short, nontechnical explanation. Code and archived analysis inputs are in the separately citable replication package: https://doi.org/10.5281/zenodo.22880953 (latest-version concept DOI). Its included data reproduce the main analyses, while raw ACRIS/DOF record linkage is supplied for provenance and was not re-tested from that deposit; the separate raw extract used to measure the condominium burden is not included there. This is an independent working paper, not a journal-accepted article.