Procurement Concentration and Potentially Recoverable Competitive Savings in New York City Contracting

New York City registers roughly $42.4 billion in new contracts annually, of which about $21.6 billion (50.9%) is awarded through limited or non-competitive methods rather than open bidding. This paper narrows that headline to an addressable base of the Sole Source and Negotiated Acquisition subgroups (about $5.66 billion, or $6.98 billion including Emergency Procurement), after excluding categories that are already competitively bid in substance or structurally non-competitive by design. It measures vendor concentration in the two largest addressable subgroups using the same Gini and top-N method independently reproduced across the AllocationAsymmetry MTA audit, and cross-checks the result against that converged benchmark. The central finding is structural, not accusatory: NYC’s Sole Source subgroup is nominally more concentrated than MTA’s entire procurement book (Gini 0.956 versus 0.931), but the concentration is driven overwhelmingly by two designated single-vehicle instrumentalities (NYEDC at 82.5% of Sole Source, HANYC at 24.0% of Negotiated Acquisition) that behave structurally like the already-excluded Government-to-Government category rather than private-market vendor capture. Applying literature-grounded competitive-savings rates (10 to 20%) yields a conditional planning range with a headline near $500 million to $1.4 billion per year, a NYCEDC-excluded floor near $434 million, and an intermediary-screened sensitivity of about $341 million to $682 million once both named vehicles are removed. That range is explicitly a conditional, pilot-gated planning estimate, not a verified recoverable amount. Its load-bearing assumption, that a meaningful share of this spend has viable alternative vendors, is the very thing the paper does not claim to have verified and proposes a bounded 12 to 18 month pilot to test. The paper measures concentration and reproduces arithmetic; it does not judge whether any award should have been competitively bid or whether any vendor relationship is improper. Every figure is reproducible from public NYC Comptroller (Appendix 1) and New York State (data.ny.gov) sources via the companion dataset. The paper converged under BayyinahEnterprise Audit Framework v3.0, Standing Rule 7, after three independent cross-instance rounds (F-PC-R1 non pass, then F-PC-R2 and F-PC-R3 qualifying passes; consecutive qualifying-ass count two; rotation closed by a member new to the second pass). Convergence certifies structural honesty, reproducibility, and accurate disclosure; it does not certify that the savings are recoverable or policy-ready. Audit trail, reproduction package, and full round history: DOI 10.5281/zenodo.21447219