K-8 Charter School in Wickenburg, Arizona: A USDA Community Facilities Direct Loan Model Study

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

Project summary

The credit question

The credit decision on a charter school building is not really about the building. Arizona pays a charter school per student, almost entirely from state formula aid, and the authorizer can revoke or decline to renew the charter. A lender must be convinced of three things, in this order.

First, that the students exist. The modelled project assumes about 500 students in a town whose median age is 65.2 and where only 754 residents are under 18 [4]. Per-pupil funding is the only material revenue line, so if the trade area cannot supply the enrollment, nothing else in the file matters.

Second, that per-student revenue is what the model says. Arizona's formula layers a statutory base level, grade weights, a small-school weight that rises as enrollment falls, Group B weights, charter additional assistance and the Classroom Site Fund. Each layer moves debt capacity by tens of thousands of dollars.

Third, that the charter will still be in force when the loan is half paid. A CF direct loan can run 40 years [3]; an Arizona charter runs 15 years, renewable for 20-year periods, with reviews at five-year intervals [5]. The authorizer, not the lender, judges the performance on which repayment depends.

Location and county economy

Why Wickenburg, and why it is CF-eligible

CF essential community facility loans are not made to a city or town of more than 20,000 inhabitants in the most recent decennial census, and the facility must primarily serve rural areas [3]. Inside a county of 4.69 million people [1], the site must be under that limit and outside the Phoenix urban area. Two incorporated towns meet both tests.

Gila Bend passes eligibility but fails the market test outright: 1,816 residents, falling, cannot fill a 500-seat school. Wickenburg is more than four times its size and is growing. The growth is not where a reader would expect: the Maricopa County part of town slipped from 6,690 to 6,647 between the 2020 base and 2025, while the Yavapai County part doubled from 824 to 1,658 [1]. The modelled site is in the Maricopa portion; the Yavapai growth is inside the commuting shed. CF eligibility uses the decennial census count, which a lender must confirm, but every Census estimate for the town is below 10,000.

Households, age and income

Source: ACS 2020 to 2024 5-year table-based summary file, tables B01003, B01002, B09001, B14001, B09020, B25010 and B19013 [4].

This table is the centre of the credit file. Wickenburg is a retirement community: half its residents are 65 or older and households average 1.8 people [4]. The families who live here are few. The ACS counts 486 town residents and 1,316 district-area residents enrolled in kindergarten through grade 8, in any type of school [4]. Small-area margins are wide, but the gap between those figures and 500 seats is not.

Income also sets the CF rate. The intermediate rate needs a service-area median household income below the poverty line or not more than 100% of Arizona's nonmetropolitan median, and the poverty line rate, for a facility that does not provide direct health care, also needs a project whose primary purpose is to meet health or sanitary standards [3], which a new school lacks. The town's $67,600 median is 84.5% of the $79,964 statewide median in the same ACS series [4]; whether it is at or below the RD nonmetropolitan figure is the open question. The model assumes the market tier.

County population

The county is large and still growing, slowly: 0.8% in the year to July 2025 [1]. It matters to this asset indirectly. It sets the teacher labour market, and it drives the residential growth that would eventually reach Wickenburg, which so far appears in the Yavapai part of town [1]. The school will depend on parent drop-off and a limited bus service; the model carries a transportation line.

Market analysis

Demand drivers

Charter demand in Arizona is real but no longer growing statewide. The Joint Legislative Budget Committee (JLBC) enrollment series shows the shift plainly.

Source: JLBC FY2027 Appropriations Report, Department of Education, Table 3 [6]; shares computed.

Charters hold about a fifth of public enrollment, a fair precedent for a new charter in an unserved town. But public enrollment fell 2.4% from FY2025 to FY2026, and the budget assumes a further 1.6% decline in FY2027: 0.9% from demographic projections and 0.7% from students moving to the state's Empowerment Scholarship Account (ESA) program [6]. The Department of Education estimates that about 59% of students who newly enrolled in the universal ESA program in FY2026 were in a public school the year before [6]. The ESA competes for exactly the family a rural charter recruits, and its FY2027 budget, total awards of $1,215,678,900 for 113,602 enrollees or about $10,701 each [6], is close to modelled charter revenue per student.

The local driver is the absence of choice. The model assumes the project would be the town's only charter school, to be confirmed against the State Board for Charter Schools roster and Department of Education enrollment data. That is an advantage, but with no local precedent for how many families will switch.

Supply and the competitive set

The competition for a K-8 seat is the district's schools, ESA-funded private or home education, and schools in Surprise and Buckeye to the southeast. Wittmann and Morristown families are the most likely to use those alternatives already, so their assumed capture is lower.

Trade area and capture rates

The trade area is built from six ZIP Code Tabulation Areas (ZCTAs), grouped by distance from Wickenburg.

Source: ACS 2020 to 2024 5-year table-based summary file, tables B01003, B09001, B14001, B01002 and B19013 [4]; "Enrolled K to 8" is kindergarten plus grades 1 to 4 plus grades 5 to 8.

Small-area ACS counts carry wide margins; Yarnell's 28 students in grades 5 to 8, for example, carry a margin of error of 47 [4]. The total should be read as roughly 2,500 K-8 students, plus or minus several hundred, and it covers every type of school, not only public ones.

The capture arithmetic is then straightforward.

The statewide charter share of 20.9% of public enrollment in FY2026 [6] is a fair upper reference for capture across a whole trade area, because it already includes dense urban markets with many charter options. The 500-seat plan needs about that share of every K-8 student in all six ZIP areas, including Wittmann families who live closer to Surprise, in a town the model assumes has no charter school today. Without Wittmann the plan needs 41% of K-8 students in the remaining five ZIP areas, and break-even alone needs 35%. That is not a credible assumption for a lender to accept without evidence.

Forecast

The model carries two enrollment paths and one cross-check.

The sponsor ramp is the project as described: 300 ADM in FY2029 (the 2028 to 2029 school year), rising to 360, 410, 450 and 475 by FY2033 and holding there (95% of design capacity).

The cross-check applies capture by ring to the ACS counts: 30% of the Wickenburg ZCTA, 20% of the four secondary ZCTAs and 10% of Wittmann. That yields about 449 students, or about 426 ADM at a 95% ratio, which the sensitivity grid below puts at about 1.06x, short of 1.20x. Even those rates are generous for a new school with no local precedent; a lender would want them confirmed by intent-to-enroll forms.

The conservative case assumes the school stabilizes at 320 ADM instead, modelled as a ramp of 200, 240, 275, 300 and 320 ADM by FY2033. It stands for weaker capture or retention than the ring rates, and it is the case the resized first phase is built to survive. The difference between the paths is the whole credit story.

Site and technical

Site requirements. A 500-seat K-8 campus with a multipurpose room, play fields, parent queuing and bus loop needs about 10 acres, carried at an assumed $50,000 per acre. Water, sewer and power capacity must be confirmed before purchase; the $1.8 million site line includes off-site utility extensions.

Zoning. Arizona classifies a charter school the same as a school operated by a school district for zoning and for zoning, site plan and development fees, and a municipality or county must allow a charter school wherever its zoning cannot legally prohibit district schools [7]. Hearings and administrative reviews must be expedited, with a final determination within 30 days for an administrative review and within 90 days for a process that requires a public hearing [7]. Building codes, including life and safety codes, still apply [7]. That is a real schedule advantage. One limit binds: a charter school sponsored by the State Board for Charter Schools may not be located on the property of a school district unless the school district governing board grants this authority [5], so surplus district land is not a simple option.

Construction approach and cost benchmark. The model assumes a single-storey 40,000 sq ft building (80 sq ft per seat) at $285 per sq ft hard cost, plus $1.8 million of site work and a 6% contingency. The public benchmark is the School Facilities Division formula. Its statutory funding guidelines for FY2027 awards are 90 sq ft per K-6 student at $358.46 per sq ft and 100 sq ft per student in grades 7 and 8 at $378.44, increased by 5% for rural school districts [8]. At its January 2026 meeting the JLBC also approved a 4.8% adjustment in the cost per square foot, which first applies to projects beginning in FY2028 [8]. For 500 students in grades K to 8, the guidelines imply 46,111 sq ft and $17.59 million, or $381 per sq ft (computed). The modelled development cost excluding land, financing and start-up is $16.43 million, or $411 per sq ft on 13% less space: close to the state formula in total dollars.

Environmental and federal requirements. CF loans must comply with the environmental review rules in 7 CFR part 1970. Borrowers operate under title VI of the Civil Rights Act of 1964, and title IX of the Education Amendments of 1972 applies to education programs that receive Agency assistance. Loan funds must be kept in insured, interest-bearing accounts where possible [3].

Schedule.

Development budget and sources and uses

Development budget

All figures in this budget are model inputs and results. Interim interest assumes a 50% average draw over 11 months at 7.50% (the 7.00% bank prime rate of November 9, 2026 [9] plus 0.50%) and a 0.75% fee. Capitalized interest covers the first year's CF interest; loan funds may pay interest until the facility is self-supporting, for up to three years unless the National Office approves a longer period [3]. Initial operating expenses for a period ordinarily not exceeding one year are also an eligible loan purpose [3], but the model funds them from equity.

Sources and uses

The 10.1% equity is a model choice, included because a lender wants evidence that a first-time sponsor can raise money. An applicant contributing funds must deposit them in its construction account on or before loan closing or the start of construction, whichever comes first [3].

Assumptions register

Every figure in the model that is not traced to a numbered source is an assumption. The material ones are below.

Revenue build and operating expenses

Per-student revenue build

Arizona charter funding starts from the district base support level formula in section 15-943, with three charter-specific differences: section 15-941 does not apply, the small-school weights apply to a holder of one charter, and charter additional assistance is added to the base support level [13]. Equalization assistance for a charter is paid as a single amount based on ADM, without categorical distinctions between maintenance and operations or capital [13].

Modelled FY2029 values grow the FY2027 amounts by the assumed 2.0% a year; per-student figures are model results. The small-school weight works against intuition. For a K-8 count of 100 to 499 the small school district weight is 1.278 plus 0.0003 per student below 500; for 500 to 599 it is 1.158 plus 0.0012 per student below 600 [10]. A 475-ADM school receives $285 less per student than a 300-ADM school, so a lost student takes away about 97% of average revenue, which is why break-even is lower than a flat per-pupil model shows. The weight is not applied individually if the holder contracts with a required management company, shares an identical board with another holder, is a subsidiary of a corporation with other charter-holding subsidiaries, or holds more than one charter, and the combined ADM exceeds 600 [13]. A single-charter holder is therefore a condition of the model.

The Classroom Site Fund, funded by the 0.6-cent education sales tax and part of K-12 state land trust revenues, rose $41 to an estimated $883 per weighted pupil for FY2027, and charter schools' allocations are based on current-year ADM [14], which helps a growing school. Allocations follow Group A weighted student counts [14]; whether that count picks up the small-school weight is to be confirmed, so the model weights the fund at the standard 1.158.

Cash timing is a separate risk. A charter is funded first on an estimated student count based on actual registration, then revised to actual ADM after 40, 100 or 200 days in session, and it must revise its student count, base support level and charter additional assistance before May 15 [13]. A school that opens below its estimate gives money back in the spring of Year 1, when liquidity is thinnest.

Year-by-year revenue ramp (sponsor case)

All figures in this table are modelled. The ramp is expressed as ADM rather than grade counts because the formula pays one K-8 rate.

Operating expense structure at stabilization (FY2033, 475 ADM)

All figures in this table are modelled. Two ratios carry the credit. Instruction at 44.3% of revenue is below the 52.1% instructional spending percentage Arizona districts reported for FY2025 [12], but the two measures are not built the same way: a charter carries its building debt inside operating revenue. The facility cost share (debt service, O&M, insurance and replacement reserve) is 24.9% of revenue. Lenders and authorizers watch it closely because it cannot flex when enrollment falls, and a quarter of revenue is at the heavy end for a school this size.

Expense growth is set at or above funding growth (salaries 2.5%, O&M 3%, insurance 4%, against 2.0% funding inflation), so stabilized DSCR barely rises (1.30x to 1.31x) as the loan amortizes. Assuming expenses grow with revenue would show improving coverage that the statutory 2% inflation adjustment does not support.

Ten-year pro forma (sponsor ramp)

All figures in the pro forma are modelled. Cash flow after debt service equals NOI minus debt service, plus capitalized interest applied, minus the reserve deposit. "Other operating" combines contracted services, curriculum, federal pass-through, transportation and general costs.

This is the sponsor's best case. Year 1 shows a surplus only because the loan pays its own first-year interest. FY2031, the first P&I year, covers debt service exactly and goes negative after the reserve deposit. Ten-year cash accumulation averages $199,000 a year against $1.03 million of annual debt service: a thin margin under a 15-year charter.

Debt structure under the CF program

The credit-elsewhere test is likely to be met: a start-up charter with no operating history has no rating and no record for a bank to lend against, which is the model's view rather than a sourced fact. The rate still matters at the margin. Each 100 basis points on $16.9 million is about $169,000 a year of interest (computed), and the 10-year Treasury yield was 4.88% on November 9, 2026 [9]. That is why CF suits rural charters, and why the National Office concurrence the rule requires for an applicant under five years old [3] should be planned into the schedule.

The in-program alternative, a CF guaranteed loan, carries a 1.25% guarantee fee, a 0.50% periodic retention fee and an 80% guarantee for FY2026 [15]. At a commercial rate it would raise debt service without fixing the enrollment problem, so it is not modelled.

Debt service schedule

All figures in this table are modelled. By FY2038, ten years in and five years before a 15-year charter starting in FY2029 would end, the loan has amortized only 12.4%. If the charter were not renewed at year 15, the lender would hold a special-use school building against a balance of roughly $13.1 million at the end of FY2043.

Debt service coverage and break-even

DSCR by year

All figures in this table are modelled. DSCR is NOI divided by scheduled debt service. In FY2029 the debt service is paid from capitalized interest, so the ratio shows operating strength only. Days cash on hand is cumulative operating cash divided by daily operating expense and excludes the CF reserve.

The model treats 1.20x stabilized coverage with at least 45 days cash on hand as the practical floor for a lender; that is an assumption, not a program rule. The sponsor case clears it from FY2032 with thin liquidity; the conservative case never covers debt service. The authorizer measures financial performance too: a sponsor reviews a charter at five-year intervals using a performance framework and may revoke it if the holder fails to meet the framework's financial performance expectations [5]. The Board's current coverage and liquidity measures are a diligence item. A 1.00x third year would draw authorizer attention just as the lender's covenant is tested.

Break-even

ADM figures are model results; shares use headcount at a 95% ADM ratio against the ACS count [4]. Measured at FY2034 (stabilized, sponsor case), revenue could fall 5.2% with expenses unchanged before coverage hits 1.00x; expenses could rise 6.7% with revenue unchanged. Those are narrow cushions for a school whose revenue is set annually by the legislature and whose enrollment is untested.

Returns

A nonprofit charter holder does not distribute profit, so IRR is a notional measure here. It is still useful to a lender because it shows whether the real estate, valued on its income, recovers the capital put into it.

All figures in this table are modelled. The unlevered return of 5.3% barely clears the 4.75% loan rate, and the income value is $2.0 million below cost, so loan-to-value at completion is about 100%. A vacant K-8 building in a town of 8,300 would realise less still in a default. The levered IRR is high only because equity is small; it measures the subsidy in a 4.75% 35-year loan, not project strength.

Sensitivity and stress tests

Two-way grid: FY2034 DSCR by CF rate and stabilized ADM

Each cell uses the sponsor ramp to FY2032 (300, 360, 410, 450 ADM) and then holds the stated ADM from FY2033.

All figures in this table are modelled. Enrollment dominates. Moving 50 students changes coverage by about 0.15x to 0.30x; moving the rate 50 basis points changes it by about 0.08x. The CF rate matters to affordability, but the file is decided by the enrollment row.

Downside scenarios (500-seat building)

All figures in this table are modelled. Each scenario is plausible. B tests the statutory adjustment, which funds the lesser of 2% or the GDP price deflator [6]. D reflects a labour market where the statewide average district teacher salary rose only about 1% in FY2025 [12], which can reverse through legislative pay initiatives. E is the risk of families trying a new school for a year and returning to the district. Every downside takes the minimum DSCR below 0.90x, and none requires anything unusual.

What the evidence supports: a resized first phase

The model also sizes a building to the conservative case: 320 ADM, 26,400 sq ft (330-seat design), identical unit costs, on the same 10-acre site so a second phase can follow demonstrated demand.

All figures in this table are modelled. Phase 1 is tight in its ramp years and its liquidity dips below 20 days in FY2032, but it stabilizes at 1.40x on enrollment equal to about 13.6% of the trade area's K-8 students, survives a 15% enrollment miss or a three-year funding freeze, and holds debt service to 16.7% of revenue (facility share 23.8%).

Program compliance checklist

Risks and mitigants

Model determination

As structured, the modelled project is not financeable. The building's arithmetic is defensible: an $18.8 million campus close to the state's district-school formula in cost [8], at an assumed 4.75% CF rate, reaches 1.30x at 475 students. The students are the problem. Wickenburg's median age is 65.2 against 37.6 for the county; the district area has about 1,316 residents enrolled in kindergarten through grade 8 and the six-ZIP trade area about 2,471 [4]. Reaching 475 ADM takes about 20% of all of them, the statewide charter share, in a town the model assumes has no charter school; break-even takes 17.5%. The model's ring capture rates, already generous for a new school, give about 426 ADM and about 1.06x, and the conservative case of 320 ADM covers 0.80x with a $3.3 million ten-year cash deficit. No reserve or rate structure fixes that.

The same site and program support a smaller first phase. A 330-seat building with an $11.3 million CF loan reaches 1.40x at 320 ADM, breaks even at 286 ADM, and holds about 1.0x under a 15% enrollment miss or a three-year funding freeze. That version is financeable under 7 CFR part 1942 subpart A, subject to conditions a lender would likely set:

1. Charter in hand. An executed charter contract with the Arizona State Board for Charter Schools, with a stated term and first fiscal year, before RD obligates funds.
2. Enrollment evidence. Signed intent-to-enroll forms for at least 80% of Year 1 ADM (160 of 200), with home ZIP codes that confirm the trade-area mix, before construction starts.
3. Single-charter covenant. The holder keeps one charter and no management-company, common-board or subsidiary arrangement that would remove the small-school weight [13].
4. Ramp reserve. About $250,000 of sponsor cash held until FY2033 or two consecutive years at 1.20x, on top of the CF reserve.
5. Coverage and liquidity covenants. A 1.10x DSCR test from FY2031 and 45 days cash on hand from FY2033, with a consultant-review trigger below either.
6. Count reporting. October enrollment and 40-, 100- and 200-day ADM reports to the lender, and prompt notice of any authorizer intervention or notice of intent to revoke.
7. Phase 2 gate. Expansion to 500 seats only after two consecutive years at 95% of Phase 1 capacity with a waitlist, sized at that time on actual retention.
8. Cost certainty. A guaranteed maximum price construction contract within 5% of budget, and a sponsor commitment to fund overruns.
9. Rate and tier confirmation. The CF rate tier and quarterly rate confirmed at approval, with debt re-sized if the rate exceeds 5.25%, and National Office concurrence for a new applicant obtained [3].

With those conditions, the credit becomes what CF lending is meant to finance: an essential public facility, sized to the rural community it serves, with growth earned rather than assumed.