Rural Health Clinic in Greenbrier, Faulkner County, Arkansas: A USDA Community Facilities Direct Loan Model Study

Model study: an illustrative feasibility analysis built on public data for Faulkner County, Arkansas; not a client engagement.

Project summary

The credit question

A CF loan for a clinic has no tenant and no lease for a lender to rely on. The debt is paid from the gap between what payers reimburse per visit and what it costs to staff the rooms, and in primary care that gap is narrow. The modelled clinic earns a 12.1% EBIDA margin at stabilization, so small misses in volume or rate consume most of the cash available for debt service. A lender has to be convinced of five things, in rough order of damage:

1. RHC certification at this address. Medicare pays an RHC an all-inclusive rate (AIR) built from its allowable cost per visit, up to a per-visit limit, instead of fee-schedule amounts [4][5]. Without RHC status, the modelled 1.27 DSCR becomes negative cash flow. The model treats certification as unresolved until the clinic is open, surveyed and enrolled, so the lender carries this risk through construction.
2. Providers hired and productive. Stabilized volume of 20,575 visits needs 6.0 provider FTE on full schedules. A two-year APP vacancy pushes DSCR below 0.6.
3. A stable payer mix. Federal law now requires states to make Medicaid eligibility for expansion adults depend on 80 hours a month of work or other qualifying activity, starting no later than the first quarter that begins after December 31, 2026 [6]. Arkansas covers those adults through ARHOME. A four-point shift of visits to self-pay takes Year 4 DSCR below 1.0.
4. CF terms as modelled. Rate tier, 40-year amortization, principal deferral and the absence of grant funds all turn on population and income tests [2][3].
5. Capacity to carry the ramp. Years 1 and 2 lose $723,074 after debt service, which must be funded up front.

Location and county economy

Why Greenbrier

CF loans for essential community facilities are not made to a city or town of more than 20,000 inhabitants in the most recent decennial census [2]. Several Faulkner County towns are well under that limit on Census estimates: Greenbrier (6,645 in 2025), Vilonia (4,894), Mayflower (2,372), Wooster (1,150), Guy (798), Holland (602), Enola (334), Twin Groves (320), Mount Vernon (148) and part of Damascus. Conway, at 72,328, is not [1]. Greenbrier is the modelled location for four reasons.

• Size and growth. It is the largest eligible town and the fastest growing, up 16.05% from an April 2020 estimates base of 5,726, about 2.88% a year [1].
• Rural status. The 2020 Census placed 66,619 county residents in urban territory [7], exactly the population of the Conway urban area [8]; the other 56,879 (46.1%) are rural [7]. There is no Greenbrier urban area in the 2020 list [8], and the Conway urban area exceeds Conway city's 64,150 estimates base [1] by only about 2,500 people, so it almost certainly stops short of Greenbrier. The parcel still needs a block-level check.
• An underserved catchment. The ZIP areas north of town (Damascus 72039, Enola 72047, Guy 72061, Mount Vernon 72111) had no health care establishments listed in 2023 ZIP Code Business Patterns [9].
• Access. Greenbrier sits on the US-65 corridor that northern residents already drive.

The trade-off is priority scoring. A project in a rural community of 5,500 or fewer earns 20 points, and of 2,500 or fewer 25 points, under 7 CFR 1942.17(c)(2)(iii)(A) [2]. Greenbrier's April 2020 estimates base of 5,726 [1] suggests it is above 5,500 and earns none, while Vilonia would earn 20; the decennial count governs. Points rank eligible applications when the State Director selects which to process [2]; they do not decide eligibility. The demand case favours Greenbrier.

County indicators

Two facts carry credit weight. The older population is growing fastest, up 21.62% since the April 2020 base [10], which supports volume and the cost-based Medicare share. County income is 115% of the state figure [11], which, because CF rate tiers and grant percentages are income-tested [2][3], is the main reason this model assumes the market rate and no grant. The model also assumes steady competition for medical assistants and front-desk staff, reflected in a 3% wage escalator.

Employment base

Health care is the largest private sector, with 17.9% of jobs [12]. Three hospital establishments employ 1,592 people, while offices of physicians employ only 852 at 78 establishments county-wide [12]. The labour pool is real, but much of it sits with a few hospital employers that also compete for the same clinicians.

Market analysis

Service area and population

The primary service area (PSA) is five ZIP Code Tabulation Areas (ZCTAs) along and beside US-65 north of Conway. North Conway and Van Buren County are secondary.

Source: ACS 2024 5-year table-based summary file, tables B01003, B09001, B09020 and B19013 [13]. Margins are wide in the small ZCTAs; the Guy population estimate of 207 carries a margin of 155, too wide to use alone [13].

The PSA holds about 23,247 people, 3,935 of them 65 or older [13]. Its distinctive feature is youth: 28.7% are under 18, against 22.6% in the county and the state in the 2024 1-year survey [11]. The model expects that pediatric volume to be paid largely by Medicaid (ARKids) and family commercial plans. The northern ZCTAs are older (35.0% of Enola's residents and 38.2% of Guy's are 65 or older), and Mount Vernon has the lowest median household income in the PSA, $52,059 [13]. These are the residents farthest from Conway's clinics.

Insurance coverage

About one in eleven county residents under 65 is uninsured, rising to 17.3% at or below 138% of poverty, the income band of the Medicaid expansion [14]. That group is the most exposed to the new federal community engagement rule. Public Law 119-21 requires states to make Medicaid eligibility for expansion adults, and for equivalent coverage under a waiver, depend on at least 80 hours a month of work, community service or a work program (or a combination), half-time enrollment in education, or monthly earnings of at least the federal minimum wage times 80 hours. The requirement starts no later than the first day of the first quarter that begins after December 31, 2026 [6]. Arkansas covers its expansion adults through ARHOME. The direction for payer mix is clear: some ARHOME volume will become self-pay, and the model tests that.

Shortage-area status

An RHC must be located in a rural area that is designated as a shortage area [15]. Under the same regulation, a rural area is one not delineated as an urbanized area by the Census Bureau, and a shortage area is one the Department of Health and Human Services designates as short of personal health services or of primary medical care manpower [15].

The designation that covers the site, its type and its date must be confirmed in writing with HRSA or the Arkansas Primary Care Office before commitment. This model treats that confirmation as a condition, not a fact. If no qualifying designation covers the site, the borrower must pursue a new designation application, which the lower-income northern ZCTAs may support, or accept fee-schedule economics that do not service the debt. How CMS applies the 2020 Census urban areas to the regulation's term "urbanized area" is a related diligence item.

Existing supply

The health care category also counts dentists, therapists, home health and social assistance. The PSA has four physician offices, all in Greenbrier, against 68 in Conway's two ZIP codes [9]: a commuter catchment whose residents get care where they work. The model includes no planned clinics, so new competition is treated as a risk, not a base-case input. A hospital-affiliated clinic in Greenbrier would be the most damaging entrant, competing for the same commercial patients and clinicians.

Demand and capture

The visit rate is a single all-ages assumption. A young PSA may use fewer physician visits per person, while visits delivered by nurse practitioners and physician assistants add volume that physician-only rates miss. The model treats the two effects as roughly offsetting.

A 33.4% stabilized capture is the model's most aggressive assumption: one in three primary care visits by PSA residents would move to the new clinic. It is plausible rather than certain, given four physician offices for about 23,000 people [9][13] and growth of 1.6% a year in the county and about 2.9% in Greenbrier [10][1]. Capture is the second-largest risk after certification: the sensitivity grid shows a 5% volume shortfall takes Year 4 DSCR below 1.0.

Forecast

Productivity ramps to 70%, 88%, 97% and 100% of standard over Years 1 to 4, while provider FTE rise from 4.0 to 6.0. After Year 4, visits grow 1% a year with no added FTE, below county population growth. Medicare's share stays at 26% for ten years, although the faster-growing 65+ population would likely raise it; that is conservative because Medicare is the best-paying class in the stabilized years.

Site and technical

Site. About 3 acres near US-65 with public water and sewer, about 60 parking spaces, an ambulance bay and room to add 3,000 sq ft. RD adds a constraint a bank would not: under 7 CFR 1942.17(f)(5), the CF rate rises two percentage points if the project involves the use of, or construction on, prime or unique farmland [2]. Open land at the edge of town may carry that classification, so an in-town commercial parcel avoids both the penalty, worth about 0.20 of Year 4 DSCR in the model, and a rezoning.

Zoning and approvals. Greenbrier has an O-1 Office and Quiet Business district and C-1, C-2 and C-3 commercial districts [16]; the ordinance's use tables must confirm where a clinic is permitted. Variances, special uses and rezonings go to the Planning Commission, which meets the third Monday of each month. The application must show existing land uses within 200 feet and carries a $75 filing fee [16]. For commercial buildings, Arkansas adopts the International Energy Conservation Code [16]. A CF health care loan also needs a statement from the responsible State agency that the facility is not inconsistent with the State Medical Facilities Plan, and a statement that it meets the standards in 7 CFR 1942.18(d)(4) [2]. CF loans must also comply with the environmental review rules in 7 CFR part 1970 [2]. The on-site lab must operate under 42 CFR part 493, because an RHC provides basic tests on site, including urine chemistry, blood glucose and pregnancy tests [15].

Construction. A single-storey building on a slab, design-bid-build: about 24 exam rooms, two procedure rooms, lab, x-ray, three behavioral health offices, care coordination, staff space and a community room. Hard cost is assumed at $385 per sq ft, or $411.95 with a 7% contingency. An architect's estimate is a condition.

The borrower may ask in writing before closing for the lower of the approval-date or closing-date rate [2]; on a 12-month build that request costs nothing.

Development budget and sources and uses

Development budget

All figures in this budget are model inputs and results. All-in cost is $728 per sq ft; real estate alone (land, site, building, contingency, design) is $6,033,000. The ramp reserve is sized from the model: $723,074 of Year 1 and 2 losses after debt service plus 90 days of Year 1 operating expense ($473,105), rounded to $1,200,000. Interim interest assumes a 55% average draw on $5,250,000 at 7.5% for 12 months.

Sources and uses

The model sizes the loan by cash flow. The binding program rules are credit elsewhere (the applicant certifies in writing that it cannot finance the project from its own resources or through commercial credit at reasonable rates and terms) and security (the as-developed market value, set by a qualified appraiser, must equal or exceed the loan plus any other debt against the security) [2]. At 87.0% of real estate cost, the loan meets the security test if the clinic appraises near cost. The 39.9% equity is what debt capacity leaves; for a mid-sized nonprofit it means a capital campaign or unrestricted reserves.

The regulation also lets loan funds pay initial operating expenses for a period ordinarily not exceeding one year when the borrower cannot pay them, and interest until the facility is self-supporting, for up to three years unless the National Office approves longer [2]. The model funds the ramp reserve with equity instead. That is the conservative case; a lender could finance part of the reserve if the letter of conditions allows it.

No CF grant. CF grant percentages require the service area's median household income to be below the higher of the poverty line or 60%, 70%, 80% or 90% of the State nonmetropolitan median, with larger shares for smaller communities, and grant funds cannot pay any cost of a project whose service-area income is above that threshold [3]. The regulation's tiers are 75% of eligible cost for communities of 5,000 or less, 55% for 12,000 or less, 35% for 20,000 or less and 15% for 50,000 or less, although the program defines a rural area as one of 20,000 inhabitants or less [3]. The model treats the RD National Office's Arkansas nonmetropolitan median as unknown and bounds the test instead. Greenbrier's $57,214 [13] fails even the 90% tier unless that median exceeds $63,571, which is above the $60,773 median of all Arkansas households in the same ACS 5-year series [13]. The Greenbrier ZCTA ($76,463 [13]) fails by more. A grant is upside only if the State Office documents a lower-income service area.

Revenue build and operating expenses

Assumptions

How the clinic gets paid

Medicare pays an independent RHC an all-inclusive rate for each visit, subject to a per-visit payment limit [4]. The Medicare Administrative Contractor sets the rate at the start of each cost reporting period by dividing estimated allowable costs by estimated visits, and reviews and adjusts it during the period [5]. Allowable costs follow Medicare's provider cost principles for items such as interest and depreciation [18]. The limit is $165 in 2026, $178 in 2027 and $190 in 2028, then rises with the Medicare Economic Index (MEI) [4]. Once the Part B deductible is met, Medicare pays 80% of the rate [4]. Two consequences run through the model.

• The limit binds only in Year 1. At low Year 1 volume, cost per visit is $234.05 and Medicare pays the $190 limit. From Year 2, cost falls below the limit and Medicare pays cost: $165.53 in Year 4 against a $203.41 limit. Medicare covers its share of cost, including depreciation and interest, but adds little margin. The margin comes from commercial and ARHOME patients.
• Productivity screens. Medicare's screening guidelines can test staffing and the cost of practitioner services [18]. The model applies a combined minimum of 4,200 visits per physician FTE and 2,100 per APP FTE (an assumption); a shortfall raises the divisor and lowers the AIR. In Year 1 the combined 9,258 visits are 192 short of the 9,450 expected, which has no effect because the limit already binds. From Year 2, actual visits clear the screen.

The model assumes Arkansas Medicaid pays the clinic a per-visit encounter rate, and that ARHOME and commercial plans pay negotiated rates.

Provider recruitment and productivity

The roster is APP-heavy by design. RHC rules require a nurse practitioner, physician assistant or certified nurse-midwife to be available to furnish patient care at least 50% of the time the clinic operates, and at least one physician assistant or nurse practitioner must be an employee of the clinic [15]. APPs also cost less than half as much as physicians per visit in this model ($46 against $80 of salary and benefits per visit at full productivity). The 1.5 physician FTE supply medical direction, collaboration with newer nurse practitioners, complex patients and the procedure rooms.

Recruitment shapes the ramp more than demand does. A rural clinic opening near a large system recruits against that system's pay scale and call schedules. Federal workforce programs that key on shortage-area status may help if the site is designated, but this model counts none of them. The 70% Year 1 productivity assumption reflects new panels and credentialing delays: credentialing a new site with each commercial plan takes time. A lender should ask for signed letters of intent with named clinicians, start dates and credentialing timelines before construction begins. It should also ask for evidence of locum coverage, because a vacancy costs about 3,450 visits a year while support staff remain on payroll.

Staffing, visits and rates

All figures in this table are modelled; the Year 1 limit of $190.00 is the 2028 statutory limit [4], and later limits apply the assumed 2.3% MEI growth. Cost per visit is allowable cost over visits after the productivity screen. The Year 8 dip reflects the end of seven-year equipment depreciation, which Medicare passes straight through. Net per visit is net patient revenue over visits, after the 3% collection loss.

Payer mix, Year 4

After the 3% collection loss, modelled net patient revenue is $3,207,897. The mix is an assumption with sourced context: the PSA has slightly fewer residents 65+ than Arkansas as a whole (16.9% against 17.6%) and far more children (28.7% against 23.1%) in the same ACS 5-year series [13], and the county's under-65 uninsured rate is 9.2% [14]. Commercial patients supply 44.2% of modelled gross revenue. If the incumbent system's employed practices keep them, and the clinic draws mainly Medicaid and uninsured patients, the margin is gone.

Operating expenses, Year 4

All figures in this table are modelled. Staff costs are 65.0% of revenue, with 15.2 support FTE at stabilization. Provider pay, support wages and staffing ratios are assumptions and should be replaced with local market data. The CF rule requires the need for malpractice insurance to be considered carefully for each health care facility it finances [2]. Owner costs (occupancy, insurance, reserve) are $228,380, or 6.6% of revenue. Debt service plus occupancy is $20.17 per visit, fixed for 40 years and partly reimbursed through the AIR's depreciation and interest.

Ten-year pro forma

All figures in the pro forma are modelled, in dollars. Revenue and expenses grow at almost the same pace from Year 4 to Year 10 (19.8% and 20.5%), so EBIDA margin holds between 11.0% and 12.2% from Year 3. The Year 8 NOI dip is the AIR passing through lower depreciation; the replacement reserve, not depreciation, funds equipment replacement. Cumulative cash flow after debt service over ten years is only $158,754, so the clinic pays its own way from Year 3 but builds little cushion beyond the RD reserve.

Debt structure

Rate tier. The poverty line rate, capped at 5%, needs service-area income below the poverty line or below 80% of the State nonmetropolitan median; health care facilities that provide direct care are exempt from its separate health-standard purpose test. The intermediate rate, the poverty rate plus half the gap to market and capped at 7%, needs income below the poverty line or not more than 100% of that median. The market rate is set using the Bond Buyer index as guidance. Rates are set at least for each quarter and adjusted to the nearest one-eighth of 1%, and incomes are measured with ACS 5-year data [2]. For the intermediate tier, Greenbrier's $57,214 would require a nonmetropolitan median of at least that amount, 94.1% of the $60,773 median of all Arkansas households in the same series [13], which the model treats as unlikely. The model therefore uses the market rate, and 5.50% is an assumption; the 10-year Treasury yield was 4.88% on November 9, 2026 [17]. If the State Office documented an intermediate-tier service area at 5.00%, Year 4 DSCR would rise to 1.34.

Debt sizing alternatives

All figures in this table are modelled. Total cost moves with the loan because interim costs and the ramp reserve do. Each $500,000 less debt adds 0.12 to 0.15 of DSCR but needs about $425,000 to $440,000 more equity, and no practical loan size holds 1.20 through a 5% volume miss. That points the lender toward borrower-level support rather than a smaller loan alone.

DSCR and break-even

All figures in this table are modelled. DSCR is project NOI (EBIDA) over CF debt service. The ramp reserve, not operations, covers Years 1 and 2. Year 3, at 1.10, is the year to watch: close to full productivity, amortization started, little room for a late hire.

For a nonprofit borrower, project DSCR is only half the test. A nonprofit pays no income tax and carries depreciation as a large non-cash expense. Its coverage is therefore usually measured as change in unrestricted net assets plus depreciation, amortization and interest, divided by debt service on all of its obligations. A lender would want that borrower-wide figure for three historical years and a forecast that includes this clinic's ramp losses. The reason is simple: in Years 1 and 2 the clinic draws $723,074 of cash after debt service. If the reserve is spent early, the parent's other programs must cover the gap. Restricted gifts and pledges should be excluded from coverage, because they cannot pay debt.

All figures in this table are modelled. Holding costs fixed, revenue must reach 97.4% of plan to cover debt service and 99.3% for 1.20. Volume break-even is slightly more forgiving because the cost-based AIR rises per visit when visits fall. The most important fact about this credit is that the clinic can lose only about 4% of planned visits before it stops covering debt service. Staffing is fixed, and in a rural clinic the provider is usually the reason the volume exists.

Returns

All figures in this table are modelled. These returns fail any investment test, as expected: the equity is mission capital. Three points still matter to a lender. First, the clinic valued on its own income is worth about 67% of its cost, because working capital, pre-opening and financing costs create no collateral and primary care margins are thin. Second, collateral still covers the debt: going-concern value is 1.21 times the Year 10 balance, and an appraisal near the $6,033,000 real estate cost covers the original loan 1.15 times, meeting the CF appraisal test [2]. The margin is modest for a special-purpose building, so the loan should rest on the operator. Third, the equity does not come back: levered cash flows total $1,004,345 against $3,488,938 invested, so a lender should judge the borrower's liquidity after the contribution, not before.

Sensitivity and stress tests

CF rate against volume (Year 4 DSCR)

All figures in this table are modelled. Volume dominates. Moving from 4.50% to 6.50% costs 0.25 of DSCR at plan volume, while a 5% volume miss costs 0.29 to 0.38. On a 40-year amortization, rate barely moves debt service; fixed staffing makes volume decisive. Diligence belongs on recruitment and capture, not the rate tier.

Downside scenarios

All figures in this table are modelled.

• No RHC status. With Medicare and Medicaid at assumed fee-schedule averages of $112 and $85 a visit, Year 4 revenue falls to $2,913,497 and NOI to -$130,973. Because the model treats certification as unresolved until the clinic is open and surveyed, a lender cannot remove this risk before construction. It can only reduce it with location evidence and back it with recourse.
• ARHOME contraction. A four-point shift to self-pay costs about $100,000 of Year 4 revenue and holds DSCR under 1.0 until Year 6. That is a plausible result of the federal community engagement requirement [6], so the base payer mix is not conservative.
• APP vacancy. One lost APP removes about 3,450 visits a year while support staff stay. The rule that an NP, PA or certified nurse-midwife be available at least 50% of the time the clinic operates [15] must also be met from opening.
• Prime farmland. The two-point increase [2] cuts Year 4 DSCR to 1.07; site choice removes it.
• Volume and rate stresses. At a 12% margin, primary care cannot absorb a 15% volume miss or a 10% commercial rate cut, both normal outcomes for a new clinic.
• Combined. A realistic bad case leaves Years 3 and 4 at 0.21 and 0.31, recovering only to 1.02 by Year 10.

Program compliance

Priority scoring is modest [2]: no population points (above 5,500); no income points (above 100% of the nonmetropolitan median); 5 points for health care; possibly 10 for a "truly rural" area unless the State Office treats Greenbrier as adjacent to or closely associated with Conway; and 5 or 10 for joint financing of 5% to 19% or of 20% or more. A second lender would raise the score but would take at least a parity position on shared security [2] and add debt service to a thin DSCR.

Risks and mitigants

Model determination

As structured, the modelled clinic is financeable as the obligation of an established nonprofit operator, but not as a stand-alone single-clinic borrower.

The project meets CF tests on applicant, rural location, population and purpose. It fits the 40-year term and deferral rules, and collateral covers the loan at cost. Its operating case reaches 1.27 in Year 4 and 1.46 by Year 10, which would satisfy most program lenders if the forecast were secure. It is not secure. Debt service breaks even at 95.9% of planned visits. Each of four realistic stresses takes coverage below 1.0, and loss of RHC status, which stays unresolved until the clinic is open, ends repayment capacity. A CF grant appears unavailable at the service area's income.

A lender would likely set conditions along these lines:

1. Location evidence before the letter of conditions: a census-block finding that the site is outside the Conway urban area, and written confirmation of a qualifying shortage-area designation current through the expected certification date, or a filed designation application.
2. Borrower-level support: the loan made to, or guaranteed by, an operator with at least five years of financially successful operation, which also avoids the National Office concurrence the rule requires for younger applicants [2], with a borrower-wide coverage covenant (for example 1.25x) and minimum days cash on hand (for example 90) after the equity contribution.
3. Equity in hand: the $3,488,938, including the $1,200,000 ramp reserve, deposited or committed before construction, with campaign pledges discounted.
4. Providers signed: the Year 1 roster (1.0 physician, 2.5 APP) under agreement before contract award, with the 50% APP availability rule documented [15].
5. Site: an in-town parcel in a district that permits clinic use [16], confirmed free of prime or unique farmland before the rate is set [2].
6. Reserves and reporting: the RD reserve as the letter of conditions requires [2]; quarterly visit, payer-mix and productivity reports in Years 1 to 4; timely cost reports.
7. Rate protection: a written request for the lower of the approval-date or closing-date rate [2].

With these conditions, the credit rests on the operator's balance sheet during the ramp and on the clinic's own cash flow after Year 4, a sound pattern for a rural health facility. Without them, a lender would be financing a 12% margin business with a 4% cushion on volume.