RV Park Underwriting Method: Site Nights, Seasonality and the Class's SBA Record

An RV park sells a narrow product: the right to park a recreational vehicle, or pitch a tent, on a prepared site for a night, a week or a season. The product cannot be stored. A site that stands empty on a wet Tuesday in May is revenue gone for good, and a site that could have been let twice over on a Saturday in July is still let only once. Most of the arithmetic a credit officer needs follows from those two facts. Demand has to be counted in site nights rather than in visitors. It has to be counted month by month rather than per year. And the money a park earns from guests who stay for a season or longer has to be kept apart from the money it earns from travellers, because the two carry different risks and, under the federal guarantee programs, raise different eligibility questions.

What follows is the class's method, step by step: the unit of account, a demand model built on public visitation series, the season as the period that decides the loan, the split between transient and long-term sites, the measures a lender reads, what SBA's loan file records for code 721211, the federal and state rules a park must clear before it can borrow under an SBA or USDA guarantee, and a short worked model. The class's full worked example, Worked Example: USDA Business and Industry Loan RV Park Feasibility Study in North Dakota, carries a complete cost basis, operating statement and sensitivity set, and none of that is repeated here. Two archived posts sit beside this one: Seasonality, Contract Mix, and Yield: A Modern Feasibility Model for RV Parks, on pricing and yield, and SBA and USDA Loans for RV Parks and Campgrounds: Eligibility, the Passive-Income Problem with Long-Term Sites, and the Study Both Programs Expect, on the eligibility questions in program terms. The general methods for capture, supply counts and the climb to stabilisation belong to the series articles Trade Area and Demand Methods: Capture Rates, Penetration Rates and Gravity Models, Supply and Comparable Analysis: Counting Competitors, the Pipeline and the Properties a Study Leans On and Ramp-Up and Absorption: How Long a New Project Takes to Stabilise, and How a Study Proves It.

One Industry Code, Several Kinds of Park

The federal industry classification gives the class one code. In the NAICS manual's words, 721211 holds establishments primarily engaged in "operating sites to accommodate campers and their equipment, including tents, tent trailers, travel trailers, and RVs (recreational vehicles)" ^[1]. The same entry adds that such a place may give guests access to washrooms, laundry rooms, recreation halls, playgrounds, stores and snack bars ^[2], which is why a park's revenue line rarely stops at site fees. A destination resort with paved pads, a family campground with tent loops and a cabin row, and a one-night stop beside an interstate all share the six digits.

For this class the edges of the code matter more than its centre. The manual's cross-references move the lessor of residential mobile home sites, the trailer park in its residential sense, out to 531190, Lessors of Other Real Estate Property ^[3]. Workers' camps go to a third code, 721310, beside rooming houses and dormitories, a group whose rooms may serve as a principal residence while they are occupied ^[4]. On paper, then, the classification already splits the business that sells stays to travellers from the landlord of residential pads and from the camp that houses a workforce. A real park can be all three at once, and the code a loan carries in SBA's file is whatever the file records, not a finding about how the park earns its money.

That fixes how the loan record below must be read. The Loan Analytics dataset maps code 721211 to a single asset class, and its asset map records the limitation: RV parks, RV resorts and campgrounds share one code and cannot be told apart. Every loan figure in this article is the record of all of them together, tent campgrounds and resort parks alike. How codes become classes is the subject of From NAICS Code to Asset Class: How SBA Loans Are Assigned, and Where Two Industries Share One Code. Glamping, which has no code of its own and lands in the same class, is treated in Glamping Underwriting Method: Unit Nights, Rate Evidence and a Loan Record That Shares the Campground Code.

State statutes tend to define the class by capacity rather than amenities. Minnesota's recreational camping area is any area, private or public, used by the day, the night, the week or longer to accommodate five or more tents or recreational camping vehicles, with or without a charge, subject to a list of exclusions ^[5]. Florida builds its definition of a recreational vehicle park around a place offered for pay for the parking, accommodation or rental of five or more recreational vehicles or tents ^[6]. Both definitions return later, because the license that follows from them is part of what a lender finances.

The Unit of Account: Site Nights, Not Visitors

A study of a park should keep its arithmetic in one unit from the first page to the last. Supply is the number of sites multiplied by the nights the park is open. Demand is the number of nights on which a party wants a site. Occupancy is nights sold divided by nights available, and the rate is revenue per night sold. Their product, revenue per available site night, is the one measure that lets a lender compare a park open six months with one open all year, because every closed night counts against it.

Public visitation series are kept in other units, and the conversion is where demand estimates most often go wrong. The National Park Service counts persons by the night: "A party of 2 visitors staying over for 3 nights yields 6 overnight stays" ^[7]. Its recreation visits follow a rule under which "one entrance per individual per day may be counted" ^[8], so a traveller who drives in on three days of one trip can appear as three visits. A model that reads visits as people, or overnight stays as sites, overstates the parties looking for a pitch by the size of the party and the length of the stay. The model below divides by both before it counts a single site night.

The same series measures camping inside the parks, the nearest public evidence of how much RV demand an anchor already holds. Campers in recreational vehicles and tent trailers at the Park Service's own campsites are reported as RV overnight stays ^[9], and guests of concessioner trailer courts, RV parks and tent campgrounds as concessioner campground stays ^[10]. Those counts describe nights spent inside the boundary by parties who found a site. They never record the parties turned away, and a study should not present them as unmet demand.

The in-park inventory belongs in the competitive supply, counted by type. Yellowstone has 11 campgrounds with over 2,000 established campsites, reserved in advance with one seasonal exception ^[11], yet only one of them, Fishing Bridge RV Park, offers water, sewer and 50 amp electrical hookups, and only to hard-sided vehicles ^[12]. A park outside that gate selling full hookup pads competes, inside the boundary, with one campground of its own kind, not with eleven, while a tent campground beside it competes with most of the inventory. The general supply method in Supply and Comparable Analysis: Counting Competitors, the Pipeline and the Properties a Study Leans On has to be kept by site type for this class.

Scale helps a reader judge an anchor, read with the counting rule in mind. The National Park System recorded 323,014,305 recreation visits and 13,016,577 overnight stays in 2025, with 406 parks reporting ^[13]. Yellowstone alone recorded 4,762,988 recreation visits and 1,238,983 overnight stays that year ^[14]. Set side by side, the counts show how few visits become a night inside a park, and how much of the overnight demand an anchor creates must be housed outside it.

A Demand Model Built From Anchor Visitation

The demand model for a park near a public anchor is a chain of conversions, each a ratio a study can defend or test. It has seven links.

1. Visiting persons equal the anchor's recreation visits divided by the visits each person generates.
2. Parties equal visiting persons divided by persons per party, since one party fills one site.
3. RV parties staying in the area equal parties multiplied by the share that travel with an RV and spend the night outside the anchor.
4. RV party nights equal those parties multiplied by the nights each spends in the area.
5. The subject's demand in a month equals the year's RV party nights, multiplied by the month's share of annual visits and by the subject's capture of the area's RV nights.
6. Nights sold in a month equal the lower of that month's demand and the transient sites multiplied by the month's days and by a practical ceiling on occupancy.
7. Season occupancy equals the nights sold across the open months divided by the transient sites multiplied by the open nights.

Table 1 prints every input the chain uses, each marked as assumed or sourced. The anchor is an assumption, not a named park, the monthly shares come from a published profile, and the behavioural ratios are placeholders that a real study replaces with survey and reservation evidence.

Rounded as shown; lines may not foot by one unit in the last place.

Table 2 runs the chain for the base case.

The inputs fall into three groups of different weight. The anchor's visits and its monthly profile are public: the Park Service holds monthly visitation estimates from 1979 to the present ^[15], and a study should take its anchor's own count by month rather than an annual total. Persons per party, the share travelling by RV, visits per person and nights per stay are behavioural ratios no public count supplies. A study should take them from a visitor survey, from the anchor's own published visitor research where it exists, or from the reservation records of comparable parks, and it should say which. The capture share is the study's judgment. It should start from the subject's share of the competitive sites of its own type, move up or down for location, access and what each competitor offers, and never be set as the figure that makes the loan work.

Capture is the most leveraged input. At the assumed 18 percent, the subject's demand reaches 84.5 percent of its available transient site nights before any month is capped. A few points less would empty the shoulder months long before it touched July. A study should print the capture share the loan needs beside the share the competitive count supports.

The public series carry two cautions of their own. The Park Service warns that parks change their counting methods from time to time, which can shift a trend for years after the change ^[23], and that parks sometimes fall back on estimated data after equipment failures, staffing shortfalls or events such as the COVID-19 pandemic and natural disasters ^[24]. A study that fits a trend to an anchor's visits should read the park's notes for a method change before treating a step in the series as growth. The national total moves too: the Park Service reported 323 million recreation visits in 2025, down 8.85 million, or 2.7%, from the record year of 2024 ^[25].

Other public series are context, not demand. The Bureau of Economic Analysis puts RVing at $27.5 billion of current-dollar value added in 2024, the second-largest conventional outdoor activity, with the largest state contributions in Indiana, Texas and California ^[26]. That measures economic activity across everything the account assigns to RVing, not nights, and a state's place in it says nothing about how many travellers want a site near a given anchor. A study can cite it to describe the market a park belongs to, but not to size the park's demand.

A park near an anchor rarely depends on the anchor alone. Highway traffic, events, seasonal residents and work crews can all buy site nights, and each needs its own line with its own evidence. Work crews need a separate note: the classification puts workers' camps in a different code from campgrounds ^[4], and a park whose plan rests on housing a workforce is, for the life of that contract, in a different business from the one the chain above describes.

Seasonality: The Month Decides, Not the Year

An RV park's year is a season and an off-season, and the shape of the season matters more to the loan than the year's total. The cleanest public evidence of that shape is an anchor's monthly visitation. Table 3 reproduces the profile Yellowstone publishes: average visits by month across 2014 to 2018, with each month's share of the annual total.

The profile is extreme, but its lesson holds for any seasonal park. In the demand model, May to October carry 95.95 percent of the year's visits and June to August alone carry 63.88 percent. A park open from May to the end of October has six months to earn twelve months of debt service, fixed costs and reserves, and three of those months do most of the work.

The second lesson is that capacity cannot be carried from one month to the next. Table 4 spreads the base case across the open months and caps each one at the practical ceiling.

In June, July and August the subject's demand exceeds what its transient sites can sell, and the excess is lost; in May and October the sites sit mostly empty. Across the season, demand comes to 84.5 percent of available site nights in Table 2, but occupancy after the monthly cap is 75.7 percent. Dividing a year's demand by a year's capacity overstates occupancy whenever the peak overflows, and the error grows with the peak. The cure is to model the season by month, or by week where weekends and holidays fill and midweek nights do not, and to cap each period at a ceiling the comparable parks actually reach in their own peak months. Arrivals and departures leave gaps, so no park sells every site every night.

The cap also changes how a lender should read a park's pricing. Demand turned away in July is evidence of pricing room in July, not of demand that can be moved to May. A study that raises the peak rate should show the effect on peak occupancy, and a study that expects the shoulder months to fill should show where that demand comes from, because the anchor's profile says it is thin.

Monthly counts also move from year to year, and not together. Yellowstone recorded 566,363 recreation visits in May 2025, up 8% from May 2024 and the busiest May on record ^[28], while July 2025, at 975,109, was down 2% on the year before ^[29]. A study should print several years of the anchor's monthly counts and test its model on the weakest recent season as well as the average.

Event risk sits on top of ordinary variation, and for a park built around one anchor it is concentrated. Yellowstone was shut from June 13 through June 21, 2022 by a historic flood, three entrances reopening on June 22 ^[30], and from March 24 through May 17, 2020 because of COVID-19 ^[31]. The Park Service's release on 2025 visitation notes a 43-day partial government shutdown during that year ^[32]. Each event would have cut the visitors on whom a park outside the gate depends, for reasons unconnected with that park. A study should set out the anchor's closure history, the access routes a closure would cut and the share of the subject's demand that depends on them, and it should run a closure case, as the worked model at the end of this article does.

A seasonal park's cash flow is therefore a quarterly problem as well as an annual one. USDA's Business and Industry rule says so in terms: the lender's credit evaluation includes a "Projected operational cash flow analysis on a quarterly basis for borrowers with seasonal cyclical cash flow" ^[33]. The same presentation serves any program, because an annual ratio averages a strong summer against a winter in which a closed park still pays insurance, taxes, a manager and its lender.

Conventions. This article follows the conventions of the Loan Analytics worked examples. Coverage is tested against the program floor the site's program page states, where the program has one, and against a stated 1.25x lender assumption, and the higher of the two binds. Equity is measured at the loan the cash flow supports, with fees recomputed on that loan. An acquisition is tested on the seller's historical earnings, not on projections.

Transient Sites, Long-Term Sites and Where the Line Falls

Most parks let some sites for longer than a trip. A seasonal site may be taken for the whole season by a guest who leaves a trailer in place and comes at weekends; a monthly or annual site may house someone who lives there. That revenue is steadier than transient revenue and often paid ahead, which flatters a cash flow. It also changes what the business is, and the further a park moves toward letting pads to residents, the closer it comes to the residential landlord the classification puts in another code ^[3].

Federal and state sources put the line between a stay and a residence in different places, and a study should know which line governs the question in front of it. HUD's rule describes an exempt recreational vehicle, among other conditions, as "Designed only for recreational use and not as a primary residence or for permanent occupancy" ^[34]. Table 5 sets the main lines a park can meet side by side.

The lines do not agree, and they are not meant to. Florida counts a guest registered for 6 months or less as transient; where the operator knowingly permits continuous occupancy for more than 6 months, it presumes the occupancy nontransient, a presumption that can be rebutted, and then applies the eviction procedures of part II of chapter 83 ^{[35] [36]}. A Florida park whose long-stay guests cross that line has a different remedy against one who stops paying, and the lender's collateral becomes a park partly occupied by tenants rather than guests. Florida also measures occupancy by the time a recreational vehicle is occupied by a transient guest, not the time it sits on the leased site ^[41], which describes the seasonal site exactly. Minnesota's sales tax law counts lodging and related services furnished by a campground or trailer camp as a sale ^[37], but leaves out "the renting or leasing of it for a continuous period of 30 days or more under an enforceable written agreement that may not be terminated without prior notice" ^[38].

For the study, the rule is to classify before it projects. Every site should carry a type (transient, seasonal, monthly or annual) and every reservation a length of stay, and the study should report the share of site nights and of site revenue in each band, under each threshold that bears on the loan. For an existing park the evidence is the reservation system's export, the rent roll of long-term sites with their written terms, and, where a state taxes short stays and exempts long ones, the sales tax returns. In Minnesota a qualifying long rental falls outside the definition of a taxable sale ^{[37] [38]}, so a study can set the taxable lodging a park has reported against the short-stay revenue it claims, an independent check on the mix the operator reports. For a new park the mix is a design choice, and the study should state it as one and show what happens to revenue and to eligibility if it drifts, as the worked model does.

The eligibility consequences of the mix come later. The credit consequences stand on their own. A park full of long-stay guests has high occupancy and a low ceiling: its revenue cannot rise in July, its rates reset only when terms renew, and its turnover, the source of a transient park's pricing power, has gone. A lender shown a high occupancy figure should ask how much of it is long-term before reading it as strength.

The Measures a Lender Reads, and the Evidence for Each

A credit officer reading a study of an RV park needs a short list of measures, each defined the same way in every year shown and each traced to evidence. Table 6 lists them with the rule or series that anchors each.

Beside those, a study should print the average rate per transient site night by month, revenue per available transient site night, the share of revenue from the store, laundry and other sources, the expense ratio with fixed and variable costs shown apart, net operating income per site and the loan per site. None of these has a public benchmark this series can cite without a trade source, so each must come from the subject's own records or from dated comparables, gathered as Operating Expense Evidence: Building an Expense Budget a Credit Officer Can Trace describes.

Coverage needs a definition before it needs a number. USDA's ratio puts the borrower's annual principal and interest underneath, and above it earnings before interest, taxes, depreciation and amortization, net of the replacement capital spending the borrower can reasonably expect ^[18]. A park's capital wears out on a schedule a study can predict: electrical pedestals, water and sewer laterals, pad surfaces, bathhouse plumbing. A study should therefore take a stated reserve per site off earnings before it computes coverage under any program, not only under B&I. For SBA lending, 1.15x of business coverage is the floor the site's SBA program page sets for standard 7(a) lending, citing SOP 50 10 8.1 ^[22], and the site's methodology page ties the test to whatever threshold the program and the lender set, the floor varying with program and loan size ^[44]. The worked model below binds at the higher of that stated floor and the series' 1.25x lender assumption. How the ratio is assembled, and which floor binds where, is set out in Debt Service Coverage in a Feasibility Study: How the Ratio Is Built, and Which Floor Binds.

The Class's SBA Record in the Loan Dataset

The class's SBA record is small enough to read whole. Over approval fiscal years 2017 to 2026, code 721211 drew 645 ^[45] 7(a) approvals totalling $842,833,000 ^[45]; the median 7(a) loan in the class was $821,000 ^[45]. The 504 program approved 281 ^[46] projects in those years, and their debentures sum to $240,245,000 ^[46], with a median of $576,000 ^[46] per project. A 504 dollar figure in the dataset counts only the debenture, so each project, once its first mortgage and the owner's equity are added, cost more. All of it is the record of RV parks, RV resorts and campgrounds together, because the code holds all three.

Franchising barely touches the class. Among its 7(a) approvals 8.37 percent ^[45] carry a franchise code; among the 504 projects the share is 4.63 percent ^[46], so the great majority of the record has none.

The yearly series is uneven. In fiscal 2017 the class drew 51 ^[45] 7(a) approvals, in fiscal 2021 103 ^[45] and in fiscal 2026 60 ^[45]; the 504 counts for the same three years were 13 ^[46], 45 ^[46] and 30 ^[46]. An approval record carries no motive, and nothing in it explains why fiscal 2021 stands out.

The two programs sit on different parts of the map. The 7(a) cell's list of states opens with Texas at 85 ^[45] approvals, then New York at 35 ^[45] and Michigan at 31 ^[45]. The 504 cell's opens with Minnesota at 67 ^[46] and Wisconsin at 39 ^[46], ahead of Texas at 32 ^[46]. Nothing in the file accounts for the weight of two neighbouring states in the 504 record, and this article does not guess at one. The concentration does mean that Minnesota's licensing and tax rules, read in this article, bear on a large part of the class's 504 history.

The national data pages, the 7(a) RV park loan data and the 504 RV park loan data, carry the same cells with their keys. USDA guarantees cannot be added: the public USDA record this series reads holds award totals by place with no asset class, as USDA Business and Industry Loans in Public Data: What USAspending Shows and What It Hides explains.

Reading the Class's Floors Against the Program's

The cohort floors are where the class looks least like the program around it, and they need the most care. Of the class's 7(a) approvals from fiscal 2017 to 2024, 69 ^[45] of 511 ^[45] ended cancelled, a floor of 13.50 percent ^[45]; the program-wide 7(a) floor for that cohort is 11.24 percent ^[47]. Treating approvals still waiting to disburse as cancelled lifts the class to a ceiling of 14.09 percent ^[45] and the program to 12.24 percent ^[48]. In 504 the order reverses: the class's floor is 12.16 percent ^[46] where the program's is 13.60 percent ^[49], and the ceilings are 16.22 percent ^[46] and 16.90 percent ^[50]. A cancelled approval ends before any money moves and the file gives no reason; how that ending differs from a loss is the subject of Cancellation Is Not Charge-Off: The Two Ways an SBA Approval Can End Without Repayment.

The charge-off record is the striking one. The class disbursed 274 ^[45] 7(a) loans from its fiscal 2017 to 2021 approvals, and 0 ^[45] of them had been charged off by the data date, a floor of 0.00 percent ^[45]; the program's own figure for the same years is 6.33 percent ^[51]. On the fixed horizon, which keeps a charge-off only if it came within five years of approval, the class reads 0.00 percent ^[45] against the program's 3.72 percent ^[52]. The 504 side repeats the pattern: 0 ^[46] charge-offs among 109 ^[46] disbursed projects, a floor of 0.00 percent ^[46], where every 504 loan of those years together stands at 0.35 percent ^[53].

A floor of zero describes the file to date, not the class's risk. Four limits stand behind it. First, it counts charge-offs recorded by the data date on loans approved five to ten years before, so it can only rise as the cohort ages; the argument is made in full in The Approval-Cohort Charge-Off Floor: Why a Published SBA Loss Rate Can Only Understate. Second, a disbursed loan that is still open, neither repaid nor cancelled nor written off, has its status withheld from the public file under FOIA Exemption 4 ^[54], and such open loans make up 110 ^[45] of the class's 7(a) denominator and 80 ^[46] of its 504 denominator. Third, the denominators are a few hundred loans each, enough for the dataset to print a rate but few enough that a handful of future charge-offs would move it a long way; the reasoning behind the 30-loan line is in Why a Rate on Fewer Than 30 Loans Is Withheld: Small Samples and the Suppression Rule. Fourth, the cohort holds every park in the code, so it cannot be read for a resort with paved pads, a tent campground or a park heavy with long-term sites.

Nor can a cause be read into the floors. The file does not say why the class's 7(a) cancellations run above the program's, or why its charge-offs are so far nil, and this article offers no reason. What the floors tell a lender is what the public record holds: no recorded charge-off for the class in the cohorts measured, and a cancellation share close to the program's. What they cannot do is stand in for the failure modes a study must test park by park: a short or thin season, a closure at the anchor, a drift in the stay mix, and the interest rate. Table 7 prints each figure with its basis and key.

Loan Data Behind This Article

Every figure in this table comes from the Loan Analytics loan dataset, analytics-loan-data freeze-3 (commit 360c856), and carries its key. SBA rates are approval-cohort floors and are withheld when fewer than 30 loans stand behind them.

Eligibility: The Federal Tests and the State License

Under SBA's regulation the first question for a park is whether it is an operating business or a landlord. Its passive business rule, 13 CFR 120.110(c), is aimed at "Passive businesses owned by developers and landlords that do not actively use or occupy the assets acquired or improved with the loan proceeds" ^[40]. An exception follows for Eligible Passive Companies, and it is an ownership structure, not a kind of park: the Eligible Passive Company spends the proceeds on property, buying, leasing, improving or renovating it, and leases that property to the business operating there, the Operating Company; the section permits a few other uses ^[55]. A park that sells nights to travellers, runs its office, store and bathhouses and sets its own rules is using its assets. A park whose income is mostly pad rent from residents has moved toward the landlord the rule names. The passive business rule itself sets no revenue percentage at which a park crosses that line. SBA's lending procedure is outside the sources this series quotes, and the archived post SBA and USDA Loans for RV Parks and Campgrounds: Eligibility, the Passive-Income Problem with Long-Term Sites, and the Study Both Programs Expect discusses it. Under any reading, a study can print the stay mix in revenue and in site nights, so that the lender can apply whichever test governs the loan.

The regulation also reaches the state license. Under 13 CFR 120.110(h) the bar covers activity "illegal under Federal, State, or local law" ^[56], and in the two states this article reads, running a park without its license is such an activity. Minnesota requires an annual license from its Department of Health before anyone may establish, maintain, conduct or operate a recreational camping area ^[57]. Florida requires a Department of Health permit, renewed every year, before a recreational vehicle park may be established or maintained ^[58]. The license is a condition of the loan as much as of the business, and a study should confirm it, with its site count, before it models a single night.

Three more SBA provisions bear on a park. Code 721211 carries a size standard of $10.0 million in annual receipts in 13 CFR 121.201 ^[59]. A 504 borrower's minimum contribution, administrative costs excluded, is 10 percent when no special condition applies ^{[60] [61]}. It is 15 percent where the project's building or structure is "limited or single purpose" in the rule's terms ^[60], 15 percent also for a borrower in business two years or less ^[61], and 20 percent when a business that new takes on such a property ^[60]. The regulation does not say whether a park is limited or single purpose, so a study of a 504 park should show the project at both contribution levels and say which one the CDC has applied; the 504 structure is laid out in SBA 504 Method: The Job Test, the Project Split and What a Study Must Show the CDC. Where a building (mobile homes included), machinery or equipment financed with SBA proceeds lies in a special flood hazard area, the borrower must carry flood insurance ^[43], a point for any park on a river or lake that should be settled against the flood map before the cost budget is final. Over all of it stands the credit standard of 13 CFR 120.150, under which "Loans must be so sound as to reasonably assure repayment" ^[62]. Separately, the same part lets SBA require a feasibility study among its loan conditions ^[63].

USDA's Business and Industry rule names the class outright. Its list of eligible tourist and recreation facilities includes "resort trailer parks and campgrounds operated as a public or private commercial enterprise" ^[64]. A direct owner's living quarters on such a site are allowed only once their pro rata value, worked out by square footage, has been taken out of what the loan may pay for ^[65], so an owner's house on the site has to be measured and carved out. The same rule shuts out "residential trailer parks" ^[39]. Its test is purpose: the bar reaches "residential housing where the primary purpose is independent housing", except as the tourist facility clause allows ^[39]. It also refuses property meant mainly for commercial rental when the borrower has, in the rule's words, "no control over tenants and services offered" ^[66]. Location is a separate test: the project has to sit in a rural area, cooperative and local foods loans apart ^[67], and the rule's rural test starts by leaving out every city or town whose population tops 50,000 inhabitants, with its adjoining urbanized area ^[68]. Part of that fringe can still count: where the Under Secretary of Rural Development has found a fringe area rural in character, it is not left out, and each such finding is made for the project in hand ^{[69] [70]}. Past $1,000,000.00, a guaranteed loan to a new business needs a feasibility study whose author the rule specifies: an independent qualified consultant whom the Agency finds acceptable ^[71]. Under the rule's definition, that study weighs "economic, market, technical, financial, and management feasibility" ^[72]. How a study answers each heading is set out in USDA B&I Guarantees Under 7 CFR 5001: What the Agency Tests and How a Study Answers.

The license also governs what a lender can rely on after closing. A Minnesota license states how many recreational camping sites are allowed, and that number cannot grow until expansion plans are submitted and approved by the Department of Health ^[42]. The license cannot pass to another person or place ^[73]. Florida's permit is likewise tied to its holder and its location ^[58], and a buyer must apply for a new one within 60 days after the transfer ^[74]. Three rules follow for a study. The model's site count should be the licensed count, not the count on a site plan. In Minnesota an expansion's new sites cannot be added to the license until the Department approves them, so the ramp of an expansion begins at approval, not at the end of construction. And on a purchase, the buyer's own license or permit is a closing item, because the seller's does not pass with the deed.

A Worked Mini-Example: One Season Carrying a Year of Debt

The model here is a sketch built on assumptions, and it describes no actual park. Picture 100 sites, 80 let to travellers by the night or the week and 20 let to seasonal guests for the whole season, open from May to October near an assumed anchor whose months follow the published Yellowstone profile. The debt is a 7(a) loan at prime plus a spread, amortised over 25 years; Table 1 holds every figure. The class's worked example, Worked Example: USDA Business and Industry Loan RV Park Feasibility Study in North Dakota, carries the complete version, with the project's costs, its opening years and a restructured loan.

Table 8 lays the operating year out by quarter, beginning in October, when the park closes.

Two quarters run at a loss before any debt is paid. The October quarter has one month of thin trade and the January quarter none, while fixed costs and the reserve continue, so coverage in both is below zero. By the end of March the park has paid two quarters of debt service out of cash it earned the summer before, and the cumulative line reaches its trough. The summer then repays the winter, and by the end of September the year's cash after debt service is positive. A lender reading only the annual ratio would never see the trough, and a borrower who entered the winter without that cash would miss a payment in a year whose annual coverage passed.

Table 9 runs the annual test under four changes to the base.

The base case clears by a hair, at 1.26x, and the test it clears is the 1.25x lender assumption, which outranks the site's lower 7(a) floor in this model. Every adverse case fails. Cutting demand by a fifth brings coverage to 1.07x and the loan the year can support to $2,822,947. Losing half of July to an anchor closure, an event of the kind the anchor's own history records, brings it to 1.11x; July is the month the peak overflows, so the closure takes away nights the park could actually have sold. Two more points on the rate leave coverage at 1.08x and push the winter carry to $284,939.

The long-term case is the instructive one. Turning 20 transient sites into season contracts lifts transient occupancy to 81.3 percent, which looks like strength, yet total revenue falls to $814,597 and coverage to 1.15x, because a season fee earns less than the peak nights it displaces. The long-term share of site revenue climbs to 22.3 percent. The park has become steadier and weaker at once, and its stay mix has moved toward the line the eligibility rules care about. Whenever a sponsor proposes to fill a soft season with long-term sites, a study should show this trade in numbers.

The base case also carries a winter carry of $256,670: the cash, beyond what operations produce, that the park needs to pay its debt service and cover its off-season losses between October and March. A study should print that figure, say where the cash to cover it will sit when the season ends, and size it on an adverse case rather than the base. How large a shock to run, and how to report a break-even, is the subject of Sensitivity and Stress Testing: Break-Even Points, Shock Sizes and What a Lender Should See.

Checks Before the Credit Memo

The method reduces to a set of checks a lender can run against any study of an RV park, each drawn from a section above.

• The license and its site count. The study should cite the state license or permit, the number of sites it allows and, on a purchase, the buyer's route to a license of its own, since a Minnesota license cannot be transferred ^[73] and a Florida buyer has 60 days to apply ^[74].
• The unit. Demand should be in site nights. Where it starts from a visitation count, the study should show the division by visits per person and by persons per party, since the Park Service counts persons and days rather than parties ^{[7] [8]}.
• The anchor's months. The seasonal profile should come from the anchor's own monthly counts over several years ^[15], with any method change or estimated year flagged ^[23].
• The cap. Occupancy should be modelled month by month and capped at a ceiling the comparables reach, with the peak's turned-away nights printed rather than assumed into the shoulder.
• The capture share. The capture the loan needs should sit beside the capture the competitive count supports, by site type.
• The stay mix. Site nights and site revenue should be reported by length of stay under each threshold that applies, with the evidence for each band, sales tax returns included where a state such as Minnesota taxes short stays and leaves long rentals out ^[38].
• The program fit. For SBA, evidence that the park actively uses its assets ^[40]; for B&I, evidence that it runs as a commercial enterprise rather than residential housing ^{[64] [39]}, with the owner's quarters carved out of the loan ^[65].
• Coverage after reserve, by quarter. Annual coverage struck after a stated replacement reserve ^[18], and a quarterly schedule showing the off-season trough ^[33] and the cash that carries it.
• The adverse cases. At the least: lower demand, a closure at the anchor in a peak month, a shift of sites to long-term contracts and a higher rate, each with the loan the year would support.
• The loan record, read as a floor. Any use of the class's SBA record should say that it covers every park in code 721211, that its rates are approval-cohort floors, and that a charge-off floor of zero to date forecasts nothing.

None of these checks makes a seasonal park safe. Together they make the season visible: how many nights the loan depends on, in which months, and what carries the debt through the winter in between.

Related Pages on Loan Analytics

• The Loan Analytics methodology series (all articles)
• How a feasibility study is done: the Loan Analytics methodology
• SBA feasibility studies (7(a) and 504)
• USDA feasibility studies
• Bankable feasibility studies for conventional loans
• RV Park Feasibility Study
• Loan data by state
• Loan data by asset class
• Worked examples (case studies)
• Worked Example: USDA Business and Industry Loan RV Park Feasibility Study in North Dakota
• Seasonality, Contract Mix, and Yield: A Modern Feasibility Model for RV Parks
• SBA and USDA Loans for RV Parks and Campgrounds: Eligibility, the Passive-Income Problem with Long-Term Sites, and the Study Both Programs Expect

Commission a Feasibility Study

Loan Analytics prepares lender-grade feasibility studies for SBA 7(a), SBA 504, USDA Business and Industry, conventional and EB-5 financing, built on the methods this series sets out. A study is not an appraisal: it tests whether a project's market, costs and cash flow can carry the debt a lender is asked to approve. To discuss a study for your project, contact Loan Analytics.

Daniel Smith, MAI, Loan Analytics