The park was full. It owned a third of the homes it rented.

A 100-pad manufactured housing community in a secondary Sunbelt market, under contract at a price the buyer had built from the trailing rent roll. Physical occupancy 92 percent. Gross scheduled income comfortably supporting the requested agency loan on the buyer's arithmetic. The seller had held the asset nineteen years and had spent the last six of them buying homes.

That last fact was in the offering memorandum as an achievement. Thirty of the ninety-two occupied pads carried a home the community owned and rented out, which the broker presented as a revenue enhancement: the same dirt producing $765 a month rather than $400.

It is a revenue enhancement. It is not, under this loan programme, an income enhancement, and the distance between those two words was the entire engagement.

Fannie Mae's multifamily guide sets out what counts in underwritten net cash flow for a manufactured housing community, and the operative provision is narrow and explicit. Gross rental income is built from the site rents, and the guide's own footnote states that the calculation includes the site rent for any affiliate-owned or borrower-owned manufactured home, but excludes the rent, or that portion of the rent, attributable to the manufactured home itself (1).

Read that carefully, because it is not a haircut. It is a complete exclusion of one component. The pad under a community-owned home counts at pad rent. The home on top of it counts at nothing.

Meanwhile the expenses of owning thirty homes remain firmly inside the projection. Maintenance, turnover, insurance, and eventual replacement of a depreciating chattel asset all sit in operating expense, because they are real costs the borrower will incur. The programme therefore removes the income and keeps the cost, which is the correct treatment from a lender's point of view and a brutal one from a seller's.

Two further constraints frame the same issue. The community may not exceed roughly 25 percent park-owned homes, with a stretch to 35 percent permitted where the borrower presents a business plan to reduce the share over time (2). And the underwriting applies a minimum 5 percent economic vacancy against gross potential rent regardless of what the trailing collections show (1). Sizing runs to a maximum 80 percent loan to value at a minimum 1.25 times debt service coverage (2).

At 30 of 92 occupied pads, the subject sat at roughly 33 percent park-owned. Inside the stretch, outside the standard, and squarely in the territory where the business plan is not optional.

The study modelled the subject against a comparison community with an identical pad count, identical physical occupancy, and an identical pad rent, differing only in who owned the homes.

Shared inputs: 100 pads, 92 percent physical occupancy, 92 occupied pads, pad rent $400 a month or $4,800 a year, park-owned home all-in rent $765 a month. Pad-level operating expenses at 35 percent of net rental income. Incremental home-level expense of approximately $2,000 per home for a small park-owned position and approximately $4,500 per home where the position is large enough to require dedicated maintenance capacity. Minimum 5 percent economic vacancy per the guide. Coverage at 1.25 times, priced at an illustrative 7.5 percent mortgage constant.

The countable income line is identical, which is the point worth pausing on. Both parks generate exactly $441,600 of income the agency will recognise, because both have 92 occupied pads at $400. The subject's additional thirty homes generate a further $131,400 a year of home rent that the rent roll shows, the broker capitalised, and the guide discards.

What survives is the expense. The result is roughly $1.2 million less supportable agency debt at identical physical occupancy, identical pad rent, and a rent roll that looks larger.

The same distinction between what a property collects and what a lender will recognise runs through the stabilised self-storage acquisition where physical occupancy read ninety percent and the money read seventy, and the denominator discipline is the same one applied in the workforce multifamily development tested against the income-qualified renter pool.

Nothing about the seller's strategy was irrational at the time. A vacant pad earns nothing. Placing a home the community owns fills the pad, captures the home rent premium, and in a tight market it is often the only way to fill a vacancy at all, since a resident who owns a home elsewhere will not move it: relocation runs $3,000 to $10,000 or more, which is precisely why manufactured housing turnover is so low, around 2.2 percent annually against roughly 47 percent for conventional apartments (3).

The cost arrives later and it arrives in three forms. The community becomes a landlord of depreciating personal property rather than a ground lessor of appreciating land. Expense ratios diverge sharply, with park-owned-heavy communities commonly running 55 to 65 percent against 30 to 35 percent for tenant-owned communities (3). And at disposition, the buyer pool narrows, because the agency route that prices best is the one that will not count the income.

The market context made the last point acute. National occupancy reached 94.9 percent in the second quarter of 2025, the highest in more than twenty years, with asking rents around $752 and cap rates near 5.9 percent in early 2026 (4)(5). This is a tightly bid asset class. It is bid tightly for pad rent, which is durable, inflation-linked, and cheap to service. It is not bid tightly for home rental income, which the two agencies decline to count and most banks decline to count with them (6).

The borrower had assumed the two agencies were interchangeable on this asset. They are not, and on this particular file the difference decided which door the deal went through.

Freddie Mac's competing product caps affiliate and third-party owned homes at 25 percent in aggregate, with no published stretch (9). At roughly 33 percent park-owned, the subject was outside that limit and inside Fannie's, and the allowance permitting up to 35 percent with a reduction plan is the only reason an agency execution was available at all.

The two products differ elsewhere in ways worth knowing before a route is chosen. Freddie will lend on communities as small as five pads against Fannie's fifty, which matters for the smaller rural assets that dominate the sector by count. Freddie publishes a tiered loan to value grid by term, requires a replacement reserve of at least $50 per site per year where Fannie typically requires none, and escrows taxes and insurance. Both agencies impose tenant protections on the site leases as a condition of the loan, a requirement that flows from the manufactured housing duty to serve mandate rather than from credit underwriting, and both treat those protections as documentation and monitoring obligations rather than as a haircut to income.

The broader capital context favoured the agency route regardless. Federal Housing Finance Agency multifamily purchase caps for 2026 were set at $88 billion for each enterprise, $176 billion combined, a 20.5 percent increase on the prior year, and manufactured housing sits inside the mission-driven category that each enterprise must fill to at least half its volume (10). An asset class the agencies are mandated to serve, in a year when their capacity expanded by a fifth, prices better than the alternatives. Most banks underwrite lot income only, exactly as the agencies do, so the borrower's hope that a bank would credit the home rent had no basis; a minority will, but as the exception rather than the rule (11).

The market study did the conventional work, but two elements of it are specific to this asset class and worth setting out.

Supply is hard to count. Most communities are small, privately held, and absent from commercial databases, so competing supply has to be built by inspection and telephone rather than extracted. The study inventoried every community within the trade area, recorded pad counts, pad rents, utility structure, and park-owned share where it could be established, and verified occupancy by observation rather than by asking.

Rent comparables carry adjustments that do not arise in conventional multifamily. Whether the community is on municipal water and sewer or on a private well and septic system changes both the achievable pad rent and the capital risk, and a private system is a material underwriting item in its own right. Home age across the community affects the rent the pads can carry and the community's quality rating, and the rating in turn governs agency eligibility, since the programme requires a minimum quality level and rates the community at the lowest score on any single characteristic.

The demand side used the affordability position rather than population growth. Manufactured housing competes with the bottom of the conventional rental market, so the relevant denominator is the local renter population whose income supports a pad rent plus a home payment but not a market apartment rent. Where that gap has narrowed, the community's pricing power narrows with it, which is the quiet risk in a market that has seen four consecutive years of rent growth above 5 percent.

Three findings.

The trailing rent roll overstated financeable income by roughly 23 percent. Not because anything was misreported, but because $131,400 of accurately reported annual home rent is not underwritable income under this programme. The study's first exhibit was a reconciliation between the rent roll and countable income, line by line, so that no reader could confuse the two again.

The park-owned share was inside the stretch band, which made the business plan a condition rather than a courtesy. At roughly 33 percent, the borrower could proceed under the allowance permitting up to 35 percent, but only with a documented plan to reduce the share over time. The study wrote that plan as an analytical exhibit rather than a marketing document: which homes were saleable at what price, what the local resident income distribution could support on a purchase, what financing was realistically available to a buyer of a used manufactured home given that neither agency currently purchases chattel loans at scale (7), and what the conversion would do to revenue and expense in each of the following four years.

Conversion lowers revenue before it raises value, and the projection had to show that honestly. Selling a home to its occupant removes the home rent and leaves the pad rent, so gross income falls. It also removes the maintenance, turnover, and insurance on that home, so expense falls further. Net operating income improves modestly, the expense ratio improves markedly, and the asset moves toward the composition that trades at the tighter cap rate. A projection that showed revenue growth through a conversion programme would have been wrong on its face. The one the study built showed revenue declining and value rising, which is counterintuitive and correct.

The loan resized to the pad-supportable figure, which the borrower had not modelled and the seller had not contemplated. The gap closed through a combination of a lower price, additional equity, and a seller note.

Three conditions went into the credit. A conversion schedule with annual targets and a reporting covenant, which converted the guide's business-plan allowance from a narrative into a measurable obligation. A capital reserve sized to the deferred maintenance on the thirty homes, since a community that has been buying homes for six years has usually been deferring work on the oldest of them. And an explicit modelling of the state regulatory position, because rent regulation in this asset class moved materially in 2025: Washington capped lot-rent increases at 5 percent annually effective May 2025, Oregon limited larger communities to 6 percent, and Pennsylvania advanced comparable measures (8). For an asset underwritten on pad rent escalation, a statutory cap on pad rent escalation is not a footnote.

Physical occupancy in this asset class is close to useless as a standalone credit measure, and for a specific structural reason: the sector is running at record occupancy, so nearly every file shows a full park. What differentiates one full park from another is composition, and composition is invisible in the occupancy number.

The reviewable version of the analysis is three lines. State the pad count and occupancy. State the split between tenant-owned and community-owned homes. Then reconcile the rent roll to countable income by removing the home-rent component entirely, and show the expense that remains behind. If the resulting net operating income does not support the requested loan at 1.25 times, the deal is resized, and no amount of occupancy will change that.

The corollary is worth stating for borrowers, because it inverts the usual advice. In this asset class, owning more of the asset makes the credit worse. The community that owns nothing but dirt, infrastructure, and amenities is the one the agency will lend most against.

Fannie Mae Multifamily Selling and Servicing Guide, Section 606, property income and underwritten net cash flow for manufactured housing communities, including the footnote excluding the manufactured home rent component and the minimum economic vacancy provision.

Fannie Mae Multifamily Manufactured Housing Communities term sheet, current edition: park-owned home limit and business plan allowance, minimum pad count, density guidance, quality level requirement, maximum loan to value and minimum debt service coverage ratio.

Broker and operator commentary on expense ratios for park-owned against tenant-owned communities, relocation cost, and turnover. Commercial sources with an interest in the sector, cited for direction rather than precision.

National occupancy, asking rent, and rent growth for manufactured housing communities, second quarter 2025, commercial brokerage research.

Manufactured housing community capitalisation rates and price per space, 2025 and early 2026, commercial brokerage research.

Commentary on how banks and non-agency lenders treat park-owned home income, commercial lending sources.

Federal Housing Finance Agency duty to serve reporting on the absence of enterprise chattel loan purchases, and the 2025 proposed rulemaking raising chattel expectations.

State manufactured housing lot-rent regulation enacted or advanced in 2025 in Washington, Oregon, and Pennsylvania.

Freddie Mac Optigo manufactured housing community term sheet, current edition: affiliate and third-party owned home limit, minimum pad count, loan to value and coverage grid, replacement reserve, and tenant protections.

Federal Housing Finance Agency multifamily loan purchase caps for 2026, announced November 2025, and the mission-driven definitions accompanying them.

Commercial lending commentary on bank treatment of park-owned home income.

Method note. The programme rules above are drawn from primary agency documents and are stated as rules. Rent levels, expense ratios, cap rates, and turnover figures are commercial and brokerage sources and are stated as ranges. The worked comparison applies the primary income rule to two constructed rent rolls; it demonstrates the mechanism rather than valuing any community. One figure commonly quoted in the commercial literature, an 85 percent physical and 70 percent economic occupancy threshold, does not appear in the current primary guide or term sheet and is deliberately omitted here.