Insurance Administration Costs

Overview of findings.

Index of Topics on Insurance Administration Costs

Data Sources for Baseline Spending on Insurance Administration

At the national level: spending on public and private insurance administration is published in the National Health Expenditure Accounts (NHEA), which are calculated by the Centers for Medicare & Medicaid Services. For calendar year 2016, the NHEA reported insurance administration spending in "Table 19. National Health Expenditures by Type of Expenditure and Program," under "Government Administration" (State and Federal) and "Net Cost of Private Health Insurance."

At the state level: the NHEA reports personal health care spending by state and by region, but these reports do not include insurance administration spending. Analyses of state-level single-payer legislation therefore have to rely on data collected by state agencies to estimate state-level administrative spending for both private and public insurance programs. The NIH has a listing of state agencies that collect data on healthcare here (choose your state from the menu at the top).

Six Methodologies for Estimating Insurance Administration Costs Under a Single-Payer Program

Methodology #1: Traditional Medicare Benchmark

One methodology (used by Friedman New York 2015, Friedman Rhode Island 2015, Friedman Maryland 2013, Friedman Pennsylvania 2013, Friedman National 2013) assumes that the insurance administration costs of a single-payer plan will be the same as the traditional Medicare program, which is generally 2% of total spending or lower since the 1990s.

Comments:

The administrative costs of the traditional Medicare program are reported in the annual Medicare Trustees Reports and their Supplementary Tables.

  1. A number of right-wing reports (Litow 2006, Matthews 2006, Zycher 2007) and popular columnists have claimed that the administrative costs for Medicare exclude "hidden" expenditures such as tax collection, marketing, accounting, fraud detection, and other costs. These claims are inaccurate, and all of these costs are included in the administrative costs reported in the annual Trustees Report (see Sullivan 2013).
  2. Traditional Medicare reimburses providers largely on a fee-for-service basis (per discharge for inpatient care, per day for nursing facilities, etc), which creates additional administrative expenses for processing and reviewing claims. Single-payer legislation that moves hospitals and other institutions onto global budgets would have lower administrative costs than traditional Medicare in this respect - see Methodology #3: Medicare Adjusted for Global Budgets & Fixed Costs Benchmark for more details on making this adjustment.
  3. To the extent that Medicare's administrative costs per person are fixed and do not scale with the volume of medical payments for that person, Medicare's administrative costs expressed as a percentage of total healthcare expenditures will appear artificially lower than if Medicare covered the whole population, because Medicare beneficiaries have much higher medical costs than the rest of the population, making the denominator (total healthcare spending) larger. There is very little research on the extent to which insurance administration costs (much less Medicare's administrative costs) are variable or fixed, making this challenging to evaluate. See Methodology #3 for additional discussion of adjusting for fixed costs.
  4. Traditional Medicare's administrative costs are at the low-end of the international experience with administrative costs for single-payer healthcare systems. This may be a reflection of the issue above: that some fixed costs lead countries covering an entire population (children through seniors) to spend a slightly higher percentage of total pending on insurance administration than a program like Medicare covering only seniors and people with disabilities. For 2015, for example, Medicare spent less than 1.4% of expenditures on administration, lower than every country in the OECD except for Finland, Norway, and Iceland. Most countries with tax-financed single-payer systems have insurance overhead only slightly higher than Medicare's, though (1.9% on average in 2015), while those with social health insurance schemes and non-profit insurance intermediaries generally have higher overhead costs (3.8% on average in 2015), but still much lower than the U.S. market-based system. For a more detailed discussion of the international comparative data see Methodology #4: International Spending Benchmark below.
  5. Most analyses use Medicare administrative cost data from the most recent available year only, although Medicare's administrative spending fluctuates from year-to-year. Analysts using the Medicare benchmark should average across multiple years of Medicare administrative spending data, and include highs and lows in sensitivity analyses.

Methodology #2: Medicare Plus Medicare Advantage Benchmark

RAND Oregon 2017 estimates that in 2020, a single-payer plan in Oregon would spend 6.5% of total expenditures on administrative costs, based on Medicare's administrative expenditures. The study does not include a citation, but it is likely that instead of referencing the Medicare Trusteees Annual Reports, the RAND authors are citing the National Health Expenditure Accounts (NHEA)'s estimates of Medicare's administrative costs, which include the administrative spending of private Medicare Advantage plans, which are much higher than for the traditional, publicly-administered Medicare program.

Similarly, Vermont UMass Wakely 2013 tests three different assumptions on insurance administrative costs, and their high-savings estimate is drawn from the 2010 Vermont Health Expenditure Analysis, which reports that Medicare administrative costs in Vermont were 4.8% of healthcare spending. The 2010 Expenditure Analysis reports the cost of Medicare claims for residents of Vermont "with adjustments for drugs and admin." While the analysis does not explain what adjustments it made to arrive at an estimate of 4.8% administrative spending for Vermont, it is likely that they too used total Medicare administrative costs - including Medicare Advantage overhead - and applied this higher spending ratio to Vermont costs.

Comments: Using a Medicare benchmark for administrative expenditures that includes Medicare Advantage plans doesn't make sense unless the single-payer plan being analyzed mirrors the partly privatized structure of Medicare, and will allow beneficiaries to choose privately-administered health coverage. Even under such a scenario, this benchmark assumes that exactly the same portion of residents will enroll in private plans as are currently enrolled in Medicare Advantage plans.

Methodology #3: Medicare Adjusted for Global Budgets & Fixed Costs Benchmark

Lewin Minnesota 2012, Lewin Colorado 2007, Lewin California 2005, and Lewin Vermont 2001 use traditional Medicare administrative spending as a benchmark for estimating administrative costs under a single-payer plan in Minnesota, but make adjustments for the elimination of hospital claims processing under global budgets and some fixed administrative costs. From the 2012 study:

  1. The Lewin Group first starts with the total dollar administrative cost per Medicare beneficiary, instead of the percentage of total expenditures spent on administration. Using total dollar spending assumes initially that all Medicare administrative costs are fixed (not variable) per capita.
  2. Second, Lewin then reduces this administrative spending estimate by 20% to account for the elimination of hospital claims processing under a single-payer plan in Minnesota. Medicare spending on the administration of hospital claims is not published, so the Lewin Group uses data on the share of Minnesota private insurers' administrative costs dedicated to hospital claims to make an estimate of savings under a Medicare-like plan with global budgets for institutions.
  3. Finally, Lewin further reduces this administrative spending estimate by assuming that administrative costs for Minnesotans under age 65 will be 42.8% of the cost of administration for the Medicare population (65+ and disabled), due to lower physician and other claims processing - essentially adjusting Medicare administrative spending by hand for variable costs.

Comments: This approach yielded an estimate of insurance administrative costs of about 1.6% per beneficiary under single-payer reform in Minnesota, very close to traditional Medicare's administrative spending. This methodology involves some improvements on Methodology #1, because it is intentional about adjusting for the elimination of hospital claims processing under a reform with global budgets (which will reduce estimates of administrative spending), as well as estimating variable vs. fixed administrative costs when the administration of traditional Medicare is applied to a healthier population (which will increase estimates of administrative spending). However, both adjustments involve some guesswork about a) what Medicare spends on hospital claims processing, and b) what share of Medicare's administrative costs are fixed and what share are variable. Both adjustments could be improved with further information and with reference to administrative spending in other countries (see immediately below).

Methodology #4: International Spending Benchmark

Government Accountability Office National 1991 and GAO National 1992 estimated the insurance administrative costs of adopting a "Canadian-style" healthcare system in the United States by applying the 1.2% insurance overhead costs of Ontario in 1987 to the United States, estimating a 79% decline in insurance administrative costs. Solutions for Progress Massachusetts 2000 applied the same 79% reduction to its estimates of spending under a single-payer plan in Massachusetts.

Physicians for a National Health Program National 1991 estimates that a single-payer proposal structured similarly to Canada's would share Canada's 1.4% administrative overhead in 1987, expressed as a percentage of personal healthcare expenditures.

Colorado Health Institute 2016 references Nicolle and Mathauer 2010 on insurance administrative costs around the world using WHO and OECD data, and estimates that single-payer in Colorado would have insurance overhead of 4.08% based on an analysis of Table 3 in the 2010 report. (Table 3 lists administrative overhead for private health insurance, and includes a range of administrative costs for each country. Nicolle and Mathauer themselves state that the average insurance administration across WHO and OECD countries from 2000-2007 was 4.7%, and 3.8% for high-income countries, so it's unclear how the authors arrived at the 4.08% aggregate figure.)

PERI California 2017 references Collins et al 2009 for insurance administrative costs in countries ranging from 1.9% in Finland to 5.6% in Germany, and estimates that 5% administrative costs is a conservative assumption for a single-payer plan in California. Collins et al 2009 are citing OECD data from 2005.

Vermont UMass Wakely 2013 tests three different assumptions on insurance administrative costs. Their low-savings estimate is based on a McKinsey Global Institute 2008 study comparing healthcare spending in the U.S. with 13 other OECD countries. According to the study authors, McKinsey finds that "the US spent 14% more than expected on health care administration" and so they "used this study as the basis of the low estimate of potential payer savings, by assuming that 14% of payer administrative costs would be eliminated" (page 46). This is a misreading of the McKinsey report, which states that the U.S. spends 14% more of total healthcare spending (not just administrative spending) than peer countries on healthcare administration, and that the U.S. actually spends 62.8% more on insurance administration than peer countries. Additionally, McKinsey adjusts all healthcare spending figures (including insurance administration) by the per capita income of the countries they compare, which they call "Estimated Spending According to Wealth (ESAW)." This is a questionable adjustment when comparing insurance administrative costs across developed countries.

Comments:

Data on insurance administrative spending for OECD countries can be accessed using the OECD.Stat online database. On the left menu, choose Health->Health expenditure and financing->Health expenditure and financing, then from the top menu select Customise->Selection->Function, which pops up a menu of healthcare functions to filter by: choose only "Governance and health system and financing administration," then click "View Data." When the table appears, on the top menu next to "Measure" open the drop-down menu and select "Share of current expenditure on health" to display administrative spending as a share of total health spending. (OECD's documentation of methodology and accounting standards for health administration and insurance is here.)

  • International benchmarks are a viable alternative or complement to Medicare's administrative spending rates when modeling a U.S. based single-payer plan, but studies using this benchmark often include reference countries with dissimilar financing and payment structures. Countries with tax-based financing and a single payment pool (Australia, Canada, Denmark, Finland, Iceland, Ireland, Italy, New Zealand, Norway, Poland, Spain, Sweden, and United Kingdom) clearly bunch together with the lowest administrative costs - 1.9% of total healthcare spending on average in 2015, with a minumum of 0.6% in Norway and a max of 2.9% in Spain. Countries with Bismarck-style social health insurance systems and regulated non-profit insurers (Austria, Belgium, France, Germany, Greece, Hungary, Japan, Luxembourg, Netherlands, and Switzerland) bunch together with slightly higher administrative costs - 3.8% of total healthcare spending on average in 2015, with a minimum of 2.1% in Hungary and a max of 6.0% in France. Authors should not average across the OECD or WHO, but across countries with similar administrative designs to the plan they are analyzing for the U.S.(Wagstaff 2009 helpfully organizes OECD countries by financing structure and provider payment systems in Table 1.)
  • As with other methodologies, most international benchmarks use administrative cost data from the most recent available year only, although in most countries (including the U.S.) administrative spending fluctuates from year-to-year. Analysts should develop their benchmark by averaging across multiple years of administrative spending data, and include highs and lows in sensitivity analyses.
  • There is confusion in some studies about the denominator used in calculating administrative spending rates: some sources have divided administrative spending into personal healthcare spending (which excludes overhead costs), while others have divided it into total healthcare spending (which includes overhead costs). The former will lead to slightly higher rates, while the latter is the most common reporting standard. Either standard is fine, so long as the model uses the same calculation as was used to calculate the reference ratio for other countries.

Methodology #5: Adjusted Medicaid Benchmark

LECG Massachusetts 2002 estimates insurance administration costs under single-payer by tripling the cost of current state administrative spending on Medicaid and the state employee insurance plan. While transitioning to a single-payer plan would have increased the state-insured population by four to five times, the authors tripled the state's administrative costs to reflect savings from economies of scale.

Similarly, Mathematica Maine 2002 estimates insurance administrative costs at 5.0%, slightly under Maine's Medicaid administrative costs of 6.4% at the time of the study. However, Mathematica does test the impact of lower administrative estimates in their sensitivity analysis.

Comment: The use of Medicaid as a benchmark for single-payer administrative costs is problematic for two reasons:

  1. Medicaid's administration has been substantially privatized with the introduction and growth of Medicaid Managed Care Organizations (MCOs), which administered at least some benefits for 89% of Medicaid beneficiaries by 2012 (CBO 2017). MCOs are private insurance companies, sometimes for-profit, which have higher administrative costs than publicly-administered insurance plans.
  2. Medicaid is a joint federal-state program. Administrative costs associated with enrollment and benefit management are largely incurred by the state (and contracted private insurance companies). However, the federal government provides more than half of the funds for Medicaid nationally and therefore incurs administrative costs associated with revenue collection, as well as in certain areas of plan administration such as setting prescription drug rebates and enforcing state compliance.

Similar to the use of Medicare Advantage administrative spending figures above (Methodology #2), Medicaid is an inappropriate benchmark for estimating the administrative costs of a universal healthcare proposal, unless the proposal is structured similarly to Medicaid: private administration of some but not all benefits and beneficiaries; similar division of responsibilities among the federal and state governments for financing and program administration; and projected to have a similar share of beneficiaries enrolled in privately administered plans as Medicaid for reference years. Using guesswork to reduce Medicaid administrative spending estimates is not as appropriate as looking to truly comparable programs.

Methodology #6: Adjusted Private Insurance Benchmark

Hsiao Vermont 2010 analyzes the detailed administrative costs of the largest non-profit insurer in the state, Blue Cross Blue Shield of Vermont, and projects a range of savings under a single-payer plan for each category of BCBSVT's administrative spending (for example, eliminating marketing costs, while sharply reducing consulting and other specific costs). The result is a projected 4.6% administrative overhead ratio, compared to 12.3% overhead for Blue Cross and 7.6% overhead for all insurance plans in Vermont.

Vermont UMass Wakely 2013 tests three different assumptions on insurance administrative costs. Their mid-range savings assumption is that moving to a single-payer plan would result in 7.0% insurance administrative costs, which is similar to the administrative costs of self-insured plans administered by Blue Cross Blue Shield of Vermont for large employers in the state, and the self-insured plan that CIGNA administers for the Vermont State Employees Medical Plan (reported as 7.1% and 7.3% respectively in Vermont BISCHA 2009). In its final report on financing Green Mountain Care - the universal healthcare program passed but not implemented - Vermont Agency of Administration 2014 states that it assumed insurance administrative costs under Green Mountain Care for enrollees not still covered by Medicaid will be 7% (Appendix D-1, p. 3). While the report does not provide an explanation of where this assumption comes from, it's reasonable to assume that it was taken from the UMass Wakely mid-range estimate.

Comments: Self-insured plans incur all of the administrative costs of fully insured private insurance plans except for those associated with underwriting: these include the many administrative costs that would be eliminated or reduced under a single-payer proposal, such as marketing, contracting with employers, hospital claims management, and so on. The administrative costs of private self-insured plans are therefore not appropriate for estimating single-payer administrative costs without adjustment.

The methodology of adjusting the administrative costs of existing private insurance plans in a given state is appealing because it draws on the cost history of administering benefits with the same provider network that would be covered under a universal plan. However, aside from administrative expenses that would be eliminated (such as marketing), determining how much other categories of administrative costs would decline under a universal program involves mostly guesswork.

Bibliography