Twenty-seven parcels. Fourteen Partnerships.
Across the New Haven assessor's corporate-property roster, fourteen distinct entities end their names in "LLP" rather than the more common "LLC." Together they hold twenty-seven parcels. That's a small share of the roughly 7,000 corporate-owned parcels in the city — about 0.4%. But the share is the wrong metric. The point is that any landlord who chooses an LLP structure registered out-of-state instantly becomes harder to trace than every other landlord in the dataset. The structural advantage scales — even when the head count is small.
The naming is the tell. "Nash Street New Haven LLP" doesn't own anything on Nash Street. "56-58 Avon Street LLP" doesn't only own 56-58 Avon. The partnership names are templates, picked at filing time, that reference one property among many. Look at the actual parcel ownership and the same LLP appears across unrelated streets — Bishop, Quinnipiac, Mansfield, Division — none of which the entity's name advertises.
An entity name without a filing.
The Connecticut Secretary of State's business-entity search is the standard public-records starting point for tracing a corporate landlord. Type in an LLC name and you get a filing: registered agent, principal office, annual-report status, a list of named principals or members. From there a journalist can build a graph of related entities. Our clustering algorithm uses exactly those fields to link the 419 ownership groups in this map.
That mechanism breaks when a landlord uses a foreign LLP. The entity is operating in Connecticut by virtue of holding property here, but it's filed elsewhere — typically Delaware, sometimes Wyoming or Virginia. The CT Secretary of State's search surface returns nothing useful, or at best a stub record indicating the entity registered as a foreign LLP without listing principals. The agent on file is often a commercial registered-agent service whose business is to be exactly that: an address that satisfies the in-state-agent legal requirement and reveals nothing more.
When the address is on the door, not in the file.
Look at the LLP names in the dataset. Several read like a property address has been promoted to be the entity name itself. That's a deliberate choice. It satisfies the human need for a name that "sounds local" — leases, contractor invoices, mailings — without surrendering the legal advantages of an out-of-state filing.
- 1256-258 WILLOW STREET LLP4 parcels
- 2BPS REALTY LLP4 parcels
- 3219 FOUNTAIN STREET LLP3 parcels
- 4DOBBS CROSSING ASSOCIATES LLP2 parcels
- 5WHITNEY HUMPHREY ASSOCIATES LLP2 parcels
- 6HOLO LLP2 parcels
- 7NASH STREET NEW HAVEN LLP2 parcels
- 8WESTVILLE VILLAGE MERCHANTS LLC REGISTERED LLP2 parcels
- 9SCARPA FAMILY LLP · PALLADIUM LLP · TREMBLAY FAMILY LLP · (5 others)1 each
Two things to notice. First, four of the top eight contain the word "Street," "Crossing," or "Avenue" — the entity name is a New Haven address. Second, more than half of the LLP-held parcels are in just three or four ownership clusters — the same partnership name appearing on multiple, geographically unrelated parcels. That's a single beneficial owner spreading its footprint across a single legal shell. Whether that's an investment LLP, a family vehicle, or a tax structure doesn't change the data-tool problem: you can't tell from the public record.
An empty overlay, ready when the reporting fills it.
Elm City Explorer applies algorithmic clustering by default — it walks the CT Secretary of State filings and links any LLCs that share a registered agent, a named principal, or a business address. Where that clustering doesn't reach (LLPs filed elsewhere, families using one LLC each), the tool also supports an editorial overlay: a JSON file at the repo root listing addresses that public reporting has named as belonging to a specific beneficial owner. Each entry carries a citation. The overlay is currently empty by design; it'll fill in as journalism continues to do its part.
That's the right division of labor. A graph of LLCs can show you what the algorithm sees. It cannot show you what the filings hide. The overlay is the seam where reporting gets credit and where the tool acknowledges the limits of its own clustering. Until the Connecticut Secretary of State's data infrastructure expands to capture foreign LLPs operating in the state, that seam will keep mattering.
The choice between filing reform and scraper engineering.
Of the three, the first is the durable fix. The second is the engineering. The third is what we have right now.
Sources & Methodology
- New Haven VGSI assessor records — parcel ownership and entity names. The fourteen LLP entities and their parcel counts are computed directly from this source.
- Connecticut Secretary of State business-entity search (service.ct.gov/business) — primary surface for CT-registered LLCs. The "stub record" pattern is what we observe when looking up the LLP names: minimal or no filing data returned compared to comparable LLCs.
- Delaware Department of State business search, Wyoming Secretary of State, Virginia SCC — the typical destinations for out-of-state LLP filings. These are exactly the surfaces the planned scraper extension will query.
- Elm City Explorer cluster data and curated-overlay system — described in the methodology section of every Findings piece. The 419 algorithmic ownership groups are derived from CT SOS data alone; the empty curated overlay is the seam where editorial reporting fills clustering's gaps.
Twenty-seven parcels is a small denominator. The point isn't to argue these LLPs are individually consequential; it's that the structural choice they represent — a foreign-state filing for a property held in Connecticut — is a transparency gap any future landlord can reuse, and a transparency gap that a state-level fix would close at scale. As of this writing, no such fix has been proposed.