Technical appendix C

Data sources and assumptions

This appendix identifies the data sources used in the model, lists the cost report variables on which hospital profiles are based, and documents the source and basis of each assumption.

Overview

The model combines three types of input: financial and operating data reported by hospitals to the Centers for Medicare & Medicaid Services (CMS); published figures for individual hospitals; and assumptions about payment rates, cost structure, and the effects of restructuring. Each input in this appendix is classified according to the strength of its basis.

Table C1. Classification of model inputs

Classification	Definition
Reported	Reported by the hospital in its Medicare cost report.
Published	Taken from a specific published document and table, cited in the references.
Derived	Calculated by the model from other inputs or from a published payment rule.
Benchmark	Selected to fall within ranges reported in the publications cited, but not reconciled to a specific edition or table.
Analyst judgment	Selected by the analyst without a single supporting source. These values are the highest priority for validation.

Note on benchmark values. Benchmark values were selected to fall within ranges reported in the publications cited in this appendix, but they have not been reconciled to a specific edition or table. Results for the illustrative hospitals, which rely primarily on benchmark values, indicate the direction and approximate magnitude of effects and should not be interpreted as estimates for any actual institution.

Data sources

Table C2. Data sources

Source	Content	Use in the model	Classification
CMS Hospital Provider Cost Report, fiscal year 2023¹ Selected fields from Form CMS-2552-10; one record per hospital per cost reporting period (6,109 records)	Revenue, expenses, salaries, employment, depreciation, cash and investments, Medicaid revenue and charges, uncompensated care, discharges, beds, and residents	Financial profiles for individual hospitals and the national analysis	Reported
UC Medical Centers 2023-24 Annual Financial Report² University of California, Table 3.6a	Net patient service revenue by payer for UCSF Medical Center, fiscal year ending June 30, 2024	Payer mix for CMS certification number 050454. The audited entity is larger than the hospital in this cost report ($7.36 billion in net patient revenue and 44,770 discharges, compared with $6.15 billion and 36,074), and therefore includes operations reported under other CMS certification numbers. Its payer shares are applied to the revenue reported in the cost report. Published Medi-Cal revenue includes supplemental payments, which explains why Medi-Cal payments appear to exceed the cost of care.	Published

The cost report data file is not stored in the project repository; instructions for obtaining it are provided in README.md.

Table C3. Cost report variables used in the model

Variable	Use in the model
`Provider CCN, Hospital Name, City, State Code`	Hospital identification
`CCN Facility Type`	Selection of short-term acute care and critical access hospitals; identification of cost-based payment
`Fiscal Year Begin Date, Fiscal Year End Date`	Exclusion of partial-year reports
`Less Total Operating Expense`	Operating expenses
`Net Patient Revenue`	Patient revenue; calculation of the commercial payment rate
`Total Other Income`	Other revenue, including investment income
`Total Salaries From Worksheet A`	Total salaries; scaling of template wages
`Wage-Related Costs (Core)`	Benefits rate
`FTE - Employees on Payroll`	Total employment
`Contract Labor: Direct Patient Care`	Agency labor
`Depreciation Cost`	Depreciation
`Cash on Hand and in Banks, Temporary Investments, Investments`	Days cash on hand
`Total Discharges (V + XVIII + XIX + Unknown), Total Discharges Title XVIII`	Medicare share of the cost of care
`Total Days (V + XVIII + XIX + Unknown), Total Days Title XVIII`	Medicare share when discharges are not reported
`Combined Outpatient + Inpatient Total Charges, Total Patient Revenue`	Denominator for the Medicaid share
`Medicaid Charges, Net Revenue from Medicaid, Cost To Charge Ratio`	Medicaid share and payment-to-cost ratio
`Cost of Uncompensated Care, Total Costs`	Uninsured share of the cost of care
`Number of Beds, Number of Interns and Residents (FTE), Rural Versus Urban`	Hospital category
`Type of Control`	Ownership (nonprofit, for-profit, or government)

Source: CMS Hospital Provider Cost Report, fiscal year 2023. Variable names are as they appear in the data file.

Assumptions

Table C4 lists each input that is not taken directly from a hospital's reported data, together with its value, the file in which it is recorded, its classification, and its basis.

Table C4. Model assumptions

Assumption	Value	Classification	Basis
Medicare payment-to-cost ratio, prospectively paid hospitals `hospital.yaml: cost_report_mapping`	0.84	Benchmark	Medicare margins reported by MedPAC³ and payment-to-cost ratios reported by the AHA⁴ imply that Medicare has paid roughly 82 to 88 cents per dollar of hospital cost in recent years. A single value is applied to all prospectively paid hospitals, and this assumption is the largest single source of uncertainty in hospital-level results.
Medicare payment-to-cost ratio, critical access hospitals `hospital.yaml: cost_report_mapping`	0.99	Derived	Medicare pays critical access hospitals 101 percent of reasonable cost⁷, reduced by the 2 percent Medicare sequestration⁸, or approximately 99 percent of cost.
Ratio of Medicare Advantage to traditional Medicare discharges `hospital.yaml: cost_report_mapping`	0.8 (0.4 for critical access hospitals)	Analyst judgment	The cost report identifies only traditional Medicare discharges. Medicare Advantage now covers about half of eligible beneficiaries nationally⁶, which would imply a ratio near 1.0. A lower value is used on the assumption that Medicare Advantage enrollees use fewer hospital services, and a lower value still for critical access hospitals, which are located in rural areas where Medicare Advantage enrollment is lower.
Medicaid payment-to-cost ratio, when not calculable `hospital.yaml: cost_report_mapping`	0.80	Benchmark	Within the range of Medicaid payment-to-cost ratios reported by the AHA⁴. Used only when a hospital's Medicaid revenue, charges, or cost-to-charge ratio is missing.
Uninsured payment-to-cost ratio `hospital.yaml: cost_report_mapping`	0	Derived	The cost of uncompensated care reported on the cost report is already net of payments received from patients.
Other government payment-to-cost ratio, cost report hospitals `model/hcris.py`	0.90	Analyst judgment	Has no effect on results for hospitals drawn from cost reports, because their other government share is set to zero.
Data quality criteria `hospital.yaml: cost_report_mapping`	Operating expenses of at least $5.0 million; commercial share of at least 3 percent; implied commercial rate between 0.7 and 6 times Medicare	Analyst judgment	Screens that exclude records with incomplete or implausible values. The reason for each exclusion is recorded.
Default benefits rate `hospital.yaml`	24 percent of wages	Benchmark	A typical ratio of benefits and payroll taxes to wages for hospital employees. Used for illustrative hospitals and for cost report hospitals whose reported wage-related costs are missing or fall outside 8 to 50 percent of salaries.
Agency labor cost premium `hospital.yaml: template`	1.6 times the cost of employed staff	Benchmark	Reflects the cost of agency and travel staff relative to employed staff performing the same work in recent years.
Staffing and wages by occupational group `hospital.yaml: template.workforce`	Nine groups; template for a hospital of about 300 beds	Benchmark	Rounded values for a community hospital. For hospitals drawn from cost reports, the template distribution is scaled to reported employment and salaries, so results by occupational group are indicative.
Distribution of nonlabor expenses `hospital.yaml: template.nonlabor`	Eleven categories	Benchmark	Rounded shares for drugs, devices, supplies, purchased services, and other categories. For hospitals drawn from cost reports, the shares are applied to reported nonlabor expenses.
Variable cost shares `hospital.yaml: template`	0 to 1 by cost line	Analyst judgment	The share of each cost line that varies with patient volume. Affects results only in scenarios that change volume.
Illustrative hospital: community nonprofit `hospital.yaml: archetypes.community`	Operating expenses of $499.7 million; operating margin of 2.5 percent; 200 days cash on hand	Benchmark	A 300-bed community hospital with a balanced payer mix. Payer mix and payment-to-cost ratios are benchmarks; the hospital does not represent an actual institution.
Illustrative hospital: academic medical center `hospital.yaml: archetypes.academic`	Operating expenses of $2.0 billion; operating margin of 3 percent; 220 days cash on hand	Benchmark	A large teaching hospital with high-acuity patients and research and teaching costs. Payer mix and payment-to-cost ratios are benchmarks; the hospital does not represent an actual institution.
Illustrative hospital: high-commercial suburban `hospital.yaml: archetypes.suburban`	Operating expenses of $599.7 million; operating margin of 9 percent; 280 days cash on hand	Benchmark	A hospital in an affluent service area where nearly half of care is commercially insured. Payer mix and payment-to-cost ratios are benchmarks; the hospital does not represent an actual institution.
Illustrative hospital: urban safety net `hospital.yaml: archetypes.safety_net`	Operating expenses of $699.6 million; operating margin of −1 percent; 90 days cash on hand	Benchmark	A hospital serving a large share of Medicaid and uninsured patients, with relatively little commercial revenue. Payer mix and payment-to-cost ratios are benchmarks; the hospital does not represent an actual institution.
Illustrative hospital: rural critical access `hospital.yaml: archetypes.rural_cah`	Operating expenses of $40.0 million; operating margin of 0 percent; 80 days cash on hand	Benchmark	A 25-bed critical access hospital paid by Medicare at 101 percent of cost, less sequestration. Payer mix and payment-to-cost ratios are benchmarks; the hospital does not represent an actual institution.
Liquidity covenant `hospital.yaml: finance`	60 days cash on hand	Benchmark	A common minimum liquidity requirement in hospital bond agreements.
Investment yield `hospital.yaml: finance`	4 percent	Analyst judgment	Applied to illustrative hospitals only. For cost report hospitals, investment income is included in other revenue.
Projection horizon `hospital.yaml: finance`	10 years	Analyst judgment	Long enough to capture the full ramp-up of all levers and the depletion of reserves in most scenarios.
Depreciation adjustment period `model/core.py`	5 years	Analyst judgment	The period over which reduced capital spending is reflected in depreciation expense.
Policy scenario: Commercial capped at Medicare `scenarios.yaml: policies.rates_only`	Administrative simplification of 15 percent	Analyst judgment	Commercial insurers pay Medicare rates but retain their own coverage rules and administrative processes. Medicaid and uninsured patients are unaffected.
Policy scenario: All payers at Medicare `scenarios.yaml: policies.all_payer`	Administrative simplification of 50 percent	Analyst judgment	Commercial insurers and Medicaid pay Medicare rates under a common fee schedule. Uninsured patients are unaffected.
Policy scenario: All payers, 5-year phase-in `scenarios.yaml: policies.all_payer_phased`	Administrative simplification of 50 percent; 5-year phase-in	Analyst judgment	Identical to the all-payer scenario, except that payment rates decline to Medicare levels in equal steps over five years.
Policy scenario: Single payer at Medicare `scenarios.yaml: policies.single_payer`	Administrative simplification of 100 percent; 3 percent increase in patient volume	Analyst judgment	All residents are covered, and care is paid for at Medicare rates under a single set of rules. Patient volume increases by 3 percent.
Response scenarios `scenarios.yaml: responses`	No response; Conventional turnaround; Aggressive restructuring; System redesign	Analyst judgment	Specify the level at which each lever is applied. The conventional turnaround reflects measures a hospital board would typically approve first; the aggressive restructuring applies all levers at their base levels; system redesign applies all levers at their high levels.
Lever levels, ramp-up periods, one-time costs, and turnover offsets (27 levers) `levers.yaml`	See Appendix A	Analyst judgment	Informed by published ranges where available; no lever level is taken from a specific source. Parameters within levers, including staff substitution ratios (1.1 support staff per registered nurse; 1.25 NPs or PAs per physician), the payment rate and avoidable cost share for closed services (75 and 80 percent), and the capital spending multiple (2.5), are also analyst judgments.
Sensitivity and probabilistic analysis ranges `model/analysis.py`	See Appendix B	Analyst judgment	The ranges tested in the sensitivity analysis and the variation in the Medicare payment-to-cost ratio in the probabilistic analysis.
Published payer mix, CMS certification number 050454 `overrides.yaml`	Medicare 20.9%; Medi-Cal 24.5%; commercial 53.8%; self-pay 0.8%	Published	Share of net patient service revenue by payer².
Hospital financial data `CMS cost report`	Varies by hospital	Reported	Revenue, expenses, salaries, employment, cash, and other reported values listed in Table C3¹.
Commercial payment-to-cost ratio `model/core.py; model/hcris.py`	Varies by hospital	Derived	Calculated so that baseline revenue equals the assumed operating margin or the reported net patient revenue (Appendix B, section 3). The implied commercial rate relative to Medicare can be compared with the estimates reported by RAND⁵.

Priorities for further validation

The following refinements would replace the assumptions with the greatest influence on results with published or hospital-specific data.

Medicare payment-to-cost ratio. Align the assumed value with a specific MedPAC report and table, and differentiate it by hospital category rather than applying a single value to all prospectively paid hospitals.
Medicare Advantage volume. Replace the fixed ratio of Medicare Advantage to traditional Medicare discharges with county- or hospital-level Medicare Advantage data.
Staffing and wages. Replace the template distribution with hospital payroll data by job category or with occupational wage data for the hospital's labor market.
Payer mix. Add published revenue by payer for additional hospitals to assumptions/overrides.yaml.

References

Centers for Medicare & Medicaid Services. Hospital Provider Cost Report, fiscal year 2023. Baltimore, MD: CMS. Selected fields from Form CMS-2552-10; accessed November 2026. Dataset description; data file.
University of California. UC Medical Centers 2023-24 Annual Financial Report. Presented to the Regents of the University of California, November 2024, item F9, attachment 3. Table 3.6a, "Net patient service revenue by major payor." PDF.
Medicare Payment Advisory Commission. Report to the Congress: Medicare Payment Policy. Washington, DC: MedPAC; published annually in March. Chapter on hospital inpatient and outpatient services. medpac.gov.
American Hospital Association. TrendWatch Chartbook. Chicago: AHA; published annually. aha.org.
Whaley CM, et al. Prices Paid to Hospitals by Private Health Plans. Santa Monica, CA: RAND Corporation; series of reports. rand.org.
KFF. Medicare Advantage enrollment update and key trends. San Francisco: KFF; published annually. kff.org.
42 C.F.R. § 413.70, Payment for services of a critical access hospital. eCFR.
Budget Control Act of 2011, Pub. L. No. 112-25, 125 Stat. 240. congress.gov.
Needleman J, Buerhaus P, Mattke S, Stewart M, Zelevinsky K. Nurse-staffing levels and the quality of care in hospitals. New England Journal of Medicine. 2002;346(22):1715-1722.
Aiken LH, Clarke SP, Sloane DM, Sochalski J, Silber JH. Hospital nurse staffing and patient mortality, nurse burnout, and job dissatisfaction. JAMA. 2002;288(16):1987-1993.
Aiken LH, Sloane D, Griffiths P, et al. Nursing skill mix in European hospitals: cross-sectional study of the association with mortality, patient ratings, and quality of care. BMJ Quality & Safety. 2017;26(7):559-568.

Typefaces: League Spartan and Roboto, distributed under the SIL Open Font License 1.1.

Note: This appendix is generated from the model's assumption files (assumptions/*.yaml) and source code (model/) by run.py --docs, and reflects the parameter values in effect on November 12, 2026.