Show Summary Details

Page of

date: 15 December 2019

# The Growth of Health Spending in the United States From 1776 to 2026

## Summary and Keywords

During the 18th and 19th centuries, medical spending in the United States rose slowly, on average about .25% faster than gross domestic product (GDP), and varied widely between rural and urban regions. Accumulating scientific advances caused spending to accelerate by 1910. From 1930 to 1955, rapid per-capita income growth accommodated major medical expansion while keeping the health share of GDP almost constant. During the 1950s and 1960s, prosperity and investment in research, the workforce, and hospitals caused a rapid surge in spending and consolidated a truly national health system. Excess growth rates (above GDP growth) were above +5% per year from 1966 to 1970, which would have doubled the health-sector share in fifteen years had it not moderated, falling under +3% in the 1980s, +2% in 1990s, and +1.5% since 2005. The question of when national health expenditure growth can be brought into line with GDP and made sustainable for the long run is still open. A review of historical data over three centuries forces confrontation with issues regarding what to include and how long events continue to effect national health accounting and policy. Empirical analysis at a national scale over multiple decades fails to support a position that many of the commonly discussed variables (obesity, aging, mortality rates, coinsurance) do cause significant shifts in expenditure trends. What does become clear is that there are long and variable lags before macroeconomic and technological events affect spending: three to six years for business cycles and multiple decades for major recessions, scientific discoveries, and organizational change. Health-financing mechanisms, such as employer-based health insurance, Medicare, and the Affordable Care Act (Obamacare) are seen to be both cause and effect, taking years to develop and affecting spending for decades to come.

# Introduction: Transformation of Medicine and Healthcare Financing During the 20th Century

Healthcare in the United States underwent revolutionary change during the 20th century, a fact of major importance that even a short glance at the historical record confirms. Yet a single look is not sufficient for understanding the complexity or dynamics of the transformation that led to modern, high-technology medicine, with its massive expenditure of funds; why the 19th century did not experience such forceful upheavals; or why the 21st century seems unlikely to do so. Tracing growth across all three centuries gives a picture that is both more complex and more coherent than any single set of images obtained from a few years or specific therapeutic developments (see Table 1 and Figure 1).1 Although the attention here is focused on medical expenditures relative to income, it is important to remain aware of the multiple developments occurring in the economy and of the social conditions that contributed as much or more to the transformation as medical care.

Figure 1. (a) U.S. National Health Expenditure share of GDP 1776–2016; (b) Annualized growth rate (15-year moving average).

Table 1. U.S. Health Expenditures, GDP, Population, and Employment, 1776 to 2016

NHE

GDP per

U.S. Pop

Life

%

Physicians/

Staff per

Health %

%GDP

cap($2009) (millions) Expectancy Urban 1,000 Pop Physician Employment 1776 1.9$1,468

2.6

43

< 5%

1850

2.2

$2,516 23 38 15% 1.86 0.1 0.8 1880 2.3$4,175

50

46

20%

1.77

0.5

0.8

1900

2.5

$5,356 76 49 40% 1.73 1.4 1.2 1910 2.7$6,325

92

50

46%

1.67

1.9

1.3

1920

3.2

$6,881 106 56 51% 1.31 2.7 1.5 1930 3.8$7,854

123

59

56%

1.34

3.9

1.8

1940

4.0

$9,595 133 64 57% 1.35 4.4 2.0 1950 4.3$14,381

152

68

64%

1.34

5.8

2.5

1960

5.0

$17,273 181 70 70% 1.27 7.4 3.1 1970 6.9$23,149

205

71

74%

1.36

9.2

4.6

1980

8.9

$28,388 227 74 74% 1.89 10.6 6.3 1990 12.1$35,874

250

75

75%

2.30

11.1

7.5

2000

13.3

$44,518 282 77 79% 2.58 11.7 8.2 2010 17.4$47,790

309

79

81%

2.72

12.9

10.6

2016

17.9

$51,420 324 79 82% 3.05 12.3 10.7 Sources: See note 1 in “Notes.” During the 18th and 19th centuries, health spending grew at a pace similar to or slightly above other consumption. After 1880, accumulating advances in medical science and a general increase in living standards pushed the rate of growth higher. By 1930, health spending had reached almost 4% of GDP, and then stayed at or near that level during the Depression, World War II, and the postwar years. Public investments in research, workforce training, buildings, and fixed capital structures supported a surge in spending after 1955 that was sustained and then accelerated as third-party financing expanded with the passage of Medicare and Medicaid legislation. Growth peaked around 1970, and then moderated over the next 50 years. The sudden and rapid rise in spending after mid-century was pivotal as the U.S. medical care system began taking on its modern shape, consolidating nationally and building the major academic medical centers that became its defining institutions. Filtering out transitory fluctuations due to business cycles, unsynchronized seasonal adjustments, and other noise reveals more clearly the excess growth in health spending (% growth in health expenditure minus % growth in GDP) that has raised the medical share of GDP sixfold, from less than 3% to more than 17%. Examining boundary issues (market vs. home production, which transactions to include, medicine vs. long-term care) and units of observation (local or national; days, months, years, or decades) provides context and reveals significant ambiguities and policy conflicts. A consideration of demographic changes indicates that population aging per se was not a major cause of cost increases, but also supports a projection that patients over age 65 are likely to account for more than half of all personal health expenditures by 2050. Forecast accuracy and ranges of uncertainty for estimates made by the Society of Actuaries (SOA), Congressional Budget Office (CBO), and Centers for Medicare and Medicaid Service Office of the Actuary (CMS OACT) show a methodological consensus that income and technology are the major drivers of growth and that funding trends are unsustainable over the long run, making another major transformation within the next 50 years quite likely. Documentation of data series and extrapolation methods are available online (Getzen, 2017). # Prior Research Although no previous study has quantitatively evaluated expenditure growth over a century or more, at least 14 studies have examined spans of 20 to 50 years (see Table 2).2 Frank Dickinson (1951, Dickinson & Raymond 1955) and J. R. Seale (1959) covered most of the years from 1929 to 1956 and reached similar conclusions: it was an era of relatively steady growth that matched the expansion of the economy, with increasing levels of government spending over time. This impression of stability is consistent with the judgments expressed in the late 1950s by George Stigler (1956), Elizabeth Langford (1957), and Joseph Garbarino (1959) regarding employment and prices. Rice and Cooper (1970) presented the official national health expenditure (NHE) data for 1968, with a brief review of trends by five-year periods going back to 1929. Reporting that health had risen from 3.5% to 6.6% of the gross national product (GNP) over 40 years, they attributed most of that growth to “population, prices, utilization and other factors” (including technology), and observed that recent growth was more rapid than the long-run average, that hospitals had replaced physicians as the dominant category of spending, and that public financing had increased substantially.3 Herbert Klarman (1977) incorporated additional years up to 1975 using various definitions of what constitutes an “era,” focusing on questions of policy, insurance, and value for money. His skeptical evaluation of prevalent causal explanations led Klarman to assert that spending growth is likely due to multiple and changing factors over time, and not to any single explanation, and he concluded that “expenditures in this country have moved steadily and markedly upward. Over the past decade [i.e., 1965-1975], the rate of increase has accelerated. In relation to the GNP, the trend in the United States is similar to that in most developed countries” (p. 232). Table 2: Studies of NHE Trends Spanning 20+ Years  Dickinson & Raymond (1955) 1929–1953 Seale (1959) 1929–1956 Rice & Cooper (1970) 1929–1968 Klarman (1977) 1929–1975 Weisbrod (1991) 1940–1990 Newhouse (1992) 1950–1990 Cutler, Rosen, & Vijan (2006) 1960–2000 Murphy & Topel (2006) 1950–2000 Hall & Jones (2007) 1950–2002 Chernew & Newhouse (2012) 1960–2009 Chandra, Holmes, & Skinner (2013) 1970–2012 Catlin & Cowan (2015) 1960–2013 Horenstein & Santos (2016) 1970–2007 Chen & Goldman (2016) 1960–2013 Burton Weisbrod’s (1991) influential essay stated: “The operational definition of health care—that is, on the boundaries of the insurance contract,” which makes it evident that he assumes a system very different from that of the 1930s, characterized by personal payment of medical fees.4 Joseph Newhouse (1992) argued that too much importance has been assigned to population aging, personal income, physician supply, utilization, and insurance, and he was quite reserved in delineating phases of growth or rates of technological change, noting: “Trying to attribute a residual to a specific factor is an inherently frustrating exercise, and the best I can do to support my argument that much of the residual is attributable to the new capabilities of medicine (i.e., technology)—is to buttress it with data I believe are consistent with it” (p. 11). Like Klarman, Newhouse asserted that “the real rate of increase in costs is similar across countries.” David Cutler, Allison Rosen, and Sandeep Vijan (2006) examined spending and mortality across four age groups and five decades and provided evidence that marginal productivity had fallen over time. More elaborate and formalized models of changes in life expectancy by Kevin Murphy and Robert Topel (2006) and Robert Hall and Chad Jones (2007) were calibrated with 1950–2000 data to analyze general equilibrium, macro-welfare effects. Michael Chernew and Newhouse (2012) observed a cumulative average annual excess growth of 2.2% unequally divided across seven decades, noting a reduction in the mid-1990s but concluding that “historical data do not indicate any persistent slowing in healthcare spending growth” (p. 37). A review in the Journal of Economic Literature by Amitabh Chandra and Jonathan Skinner (2012) makes a conceptual and theoretical distinction between three types of medical technology—cost-saving, sometimes effective, and expensive with uncertain effectiveness. Several recent empirical studies have attempted to make meaningful historical distinctions in NHE trends, splitting the last half-century into three, four, five, or eight eras of varying length and different starting and ending points. Chandra, Jonathan Holmes, and Skinner (2013) took significant moderation in cost growth since 2000 as a starting point, dividing 1990–2012 into four periods with divisions at 2001, 2007, and 2009. Although they remained skeptical about continued slow growth in the future, they predicted excess cost growth of GDP + 1.2% for the next two decades—less than half as much as during the prior 50 years. U.S. Department of Health and Human Services Office of the Actuary (OACT) researchers Aaron Catlin and Cathy Cowan (2015) divided 1960–2013 into eight parts, at 1965, 1973a, 1982b, 1992, 1999a, 2002b, and 2007a (using a and b to designate sub-eras).5 Alice Chen and Dana Goldman (2016) split the same span into four eras, at 1973, 1995, and 2002. Both studies link period markers to public policies and nominal growth rates, but do not refer to quantified dimensions of organization or insurance or provide analysis of temporal dynamics and lags in macroeconomic adjustment. Alex Horenstein and Manuel Santos (2016) divided 1970–2007 into three eras with divisions at 1977 and 1990, using comparative data on 10 Organisation for Economic Co-operation and Development (OECD) countries to make a case that most of the divergence in costs between the United States and other developed nations occurred during the 1978–1990 period. Differences in classification for the same years across studies may indicate that time-series dynamics and noise make it hard to define meaningful eras with lengths of fewer than fifteen years. # Data Sources Data for the 18th and 19th centuries are spotty and irregular. Peter Lindert compiled “city directories” from 1772 to 1815 that listed the occupational categories of the residents of Boston, Baltimore, Charlestown, New York, Philadelphia, and elsewhere (Lindert & Williamson, 2016). The decennial U.S. Census began collecting data on the personal occupations of respondents beginning in 1850 and continued through 1990, providing a long series that can be used as a proxy for expenditures (Carter et al., 2006). Surveys of family spending were carried out toward the end of the 19th century by the U.S. Bureau of Labor Statistics (BLS). A set of 15 representative consumer budgets was compiled, and a more representative and inclusive Consumer Expenditure Survey was used to determine budget item weights for the 1918 consumer price index (Bureau of Labor Statistics, 1920, 1924). The first comprehensive assessment of national health spending appears to have been made by Louis I. Dublin (1927, p. 767), past president of the American Statistical Association and a vice president at Metropolitan Life Insurance Company, who wrote: The best estimate of the aggregate cost of sickness in the United States is over two billion dollars a year . . .$80 per family per year for medical and health purposes . . . three and a half percent of the country’s income . . . approximately 150,000 physicians, whose average net income is a little more than $3,000 a year. (p. 767) Dublin became a member of the Committee on the Costs of Medical Care (CCMC), which estimated total U.S. national health expenditures at$3,656 million dollars, approximately $30 dollars per person, or 4% of national income in 1929, given a number of caveats and qualifications (Committee on the Costs of Medical Care, 1932, p. 14). U.S. national health accounting originated in the extensive 28-volume report of the CCMC published in 1932 (Fetter, 2006; Perkins, 1998). For the next three decades, researchers at the Federal Security Agency and its successor, the Department of Health, Education, and Welfare, constructed estimates of public and private expenditures, many of which were published in the Social Security Bulletin and eventually compiled by Barbara Cooper, Nancy Worthington, and Mary McGee (1973) in the Compendium of National Health Expenditures Data. They relied heavily on the evolving national income and product accounts (NIPA), which included medical care as one component of personal consumption expenditures after 1929, albeit using somewhat different definitions and methods. The expansion of health insurance, fiscal transfers, and differences in budget categories and fiscal years, all rendered this ad hoc process increasingly less satisfactory over time. In 1964, Louis Reed and Dorothy Rice constructed a new set of accounts with the aim of making them comprehensive, integrated, and comparable to the 1932 CCMC report by incorporating data from the Internal Revenue Service, American Hospital Association, American Medical Association, Veterans Administration, Department of Defense, National Institutes of Health, Public Health Service, and other sources (Reed & Rice, 1964). Subsequently, Reed and Ruth Hanft (1966) extended these estimates, providing comparable figures for 1950, 1955, and each year from 1960 to 1965, creating a foundation for the NHE series. Responsibility was assigned to the Office of the Actuary (OACT) after the Health Care Financing Administration was established in 1977 (subsequently renamed Centers for Medicare and Medicaid Services [CMS] in 2001), and official estimates were published each year in Health Care Financing Review, along with occasional analyses and projections of future expenditures. The quarterly Review ceased publication in 2009, and since then, NHE estimates and projections have been published regularly in Health Affairs and in the annual Medicare Trustees Report. Documentation of definitions, sources, and methods is accessible on the CMS website, along with the most recent set of estimates (Centers for Medicare and Medicaid Services, 2015, 2017a). Major revisions are undertaken periodically, the most recent being in 2010 (Centers for Medicare and Medicaid Services, 2010; Haber & Newhouse, 1991). There have also been a number of studies comparing the National Health Expenditure Accounts (NHEAs) with similar estimates from Bureau of Economic Analysis (BEA), national income and product accounts (NIPA), BLS Consumer Expenditure Surveys, and the Medical Expenditure Panel Survey (MEPS), some showing substantial differences of 10% or more.6 The NHEA series, which now has extensive data tables available for each year from 1960 to 2016, is widely used and well documented. Yet the details and complexity of the process and the availability of multiple valid alternative measures are not always fully appreciated. The various series used to trace the long-run growth of national health spending and the process of reconciling them are described in greater detail in a monograph (Getzen, 2017). # Historical Trends ## 1776–1880: Preclinical Science and Medical Nihilism George Washington, the first president of the United States, began to feel ill on Thursday, December 12, 1799. He was quite healthy, riding for miles and chopping wood the previous day, but he came down with a sore throat after riding in the cold rain. On Friday, he asked an overseer at Mount Vernon, George Rawlins, to bleed him for relief, and also sent for a doctor. Dr. James Craik arrived late that evening and bled Washington again. Repeated bleedings, hot compresses, and other efforts were of no avail. Washington’s breathing became labored, and he died on Saturday night (Mount Vernon, 2017; Wallenborn, 1997). The best medical practices of the time did not help and may actually have hastened his death. James Garfield, twentieth president of the United States, was shot in the back on July 2, 1881. A changing group of physicians scrambled for control over the treatment of the severely injured president. Attempting to remove the bullet, at least eight physicians used their hands, mostly unwashed, to probe inside the wound (Millard, 2011; Paulson, 2006; Rutkow, 2006). The inventor Alexander Graham Bell brought a metal detector to the bedside in the hopes of locating the deep-seated object. After several unsuccessful probes and surgeries, Garfield was moved by train to a seaside cottage for rest. On September 19, he died. Medical experts are divided on what Garfield’s prospects of survival were; some argue that doctors probing with dirty fingers had caused the infection that brought about his demise; others suggest that the traumatic injury alone would inevitably have caused death. Considering that he survived more than two months after being shot, it seems likely that modern medical treatment could have provided partial or full recovery within a few weeks or months. The examples of George Washington and James Garfield fit within the general assessment by historians of 18th- and 19th-century medicine in America—doctors caring for patients and struggling to display therapeutic effectiveness while being of little economic significance.7 The United States was a rural nation, and most households were engaged in agriculture. Purchasing medical care was a secondary concern, less important than obtaining adequate food, heat, and clothing, and most care was provided at home by family members rather than purchased in the market. The costs were relatively minor, and the number of physicians per 1,000 population declined slightly from 1850 to 1880 (from 1.86 to 1.77), as did the share of the workforce in health occupations, which was still less than 1% of total employment. Yet despite the lack of perceived therapeutic value, amounts spent on medicines and doctors rose because incomes rose, and because increasingly urbanized workers could seek professional care at doctors’ offices or clinics instead of at home. There is no definitive evidence that health spending was rising significantly faster than incomes, but it is likely that the national average was rising modestly, with the share increasing less than .25 percent per year over the hundred-year span. ## 1880–1910: Therapeutic Influx By 1880, more than 20% of the U.S. population lived in cities and over half depended on wages or business income (see Table 1). Major medical discoveries made during the 19th century stimulated the demand for professional care. René Laennec published his article on the use of the stethoscope for auscultation in 1819. Anesthesia with ether and chloroform was described in multiple publications around 1850. Pasteur put forth his germ theory of disease in 1858, his vaccination for anthrax using oxidized bacteria in the 1870s, and his rabies treatment in 1886. Joseph Lister’s 1867 Lancet article relied on Pasteur’s research to support using carbolic acid to achieve antiseptic surgery. Robert Koch demonstrated his techniques of bacterial culture and isolated the bacillus tuberculosis in 1882. Most of these developments came out of European centers of learning. In the still largely rural United States, day-to-day medical practice was little changed and not highly respected. Bloodletting remained the most common therapeutic procedure (Bryan, 1964). The Canadian Sir William Osler was a leading American physician at the end of the 19th century, appointed to the medical faculty at the University of Pennsylvania in 1885 and the new Johns Hopkins Hospital in 1889. His seven-volume The Principles and Practice of Medicine, published in 1892, became a standard textbook for the new scientific practice of medicine. The purchase of care was becoming a consumer staple in the cities and towns, and would reach into rural areas over the coming decades. The Eleventh Census of the United States, in 1900, recorded a rise in health occupations to more than 1% of total employment. The director of that census, Carroll D. Wright, who also served as Commissioner of Labor for Massachusetts (1873–1878) and for the United States (1885–1905), carried out large-scale surveys that reported expenditures on “sickness” to be 2% to 4% of the total consumption of industrial workers (Sobek, 2001; Stigler, 1954; Williamson, 1967; Wright, 1894). It should be noted that funeral expenses and coverage for lost wages were often included among sickness expenditures (Murray, 2007; Rubinow, 1913). These census occupational statistics, considered alongside the various consumer surveys and historical narratives, strongly suggest that the rate of growth in medical spending had begun to accelerate by 1900, though they are not sufficient to reliably establish the precise level or rate of annual increase. ## 1910–1930: Formative Years of Medical Organizations Four related organizational components shaped medicine during the early 20th century: education, regulation, specialization, and hospitalization.8 Traditional training for doctors included apprenticeship and perhaps some classical Greek and Latin. The 1910 Flexner Report called for medical students to be educated in chemistry, anatomy, and physiology; have a college degree; get hospital experience; and conduct laboratory tests (Beck, 2004; Flexner, 1910; Perkins, 1998). The report suggested emulating the Johns Hopkins University, which was following the German scientific academic program, as a model, and it recommended closing subpar “diploma mill” schools. A restrictive movement to raise the quality of practice was applied to the workforce through licensure. Nineteenth-century practice had legally been open to anyone who chose to call himself (or herself) a doctor. By 1930, state laws and medical boards had effectively limited the number of physicians in practice. Relative to population, the availability of physicians fell by a third, even as therapeutic effectiveness, and therefore demand, was growing. The constricting physician supply shifted the workload to ancillary staff. Previously limited to apprentices and a few nurses, the number of assistants per doctor tripled (see Table 1 in “Introduction: Transformation of Medicine and Healthcare Financing During the 20th Century”). Advanced scientific training and the devolution of routine tasks to assistants led more and more physicians to specialize. To become an expert in the eye, ear, hand, or stomach, a doctor could no longer also learn all there was to know about the full range of diseases or spend time wrapping bandages, traveling to visits patients who lived out in the country, compounding basic drugs, or carrying out multiple batches of routine chemical tests. Advanced training in medical specialties was overseen by a growing number of physician organizations and specialty boards. The American College of Surgery was founded in 1913, American College of Physicians (Internal Medicine) in 1915, American College of Ophthalmology in 1916, American College of Radiology in 1923, American Board of Internal Medicine in 1936, and American Board of Surgery in 1937. Concurrent with the development of graduate medical education, licensure, and specialization, hospitals began to change from being institutions that served the poor and disabled to being centers of advanced practice with concentrations of equipment and trained ancillary staff. By 1930 there were 6,613 hospitals in the United States, with 974,115 beds. William H. Lough (1935) carried out an extensive study of consumption in the U.S. economy with estimates for selected years from 1909 to 1931 indicating that medical care had risen from 2.7% to 3.7% of the aggregate total, which implied an excess growth rate of +1.5% annually over the 22-year span (see Table 3). Over the same period, health employment grew 1.8% per year faster than total employment, though the number of physicians per 1,000 population fell −1.1% per year.9 Given the ambiguities and measurement difficulties, a reasonable estimate of the average excess growth rate from 1910 to 1930 would be above +1.2% but below +1.8% per year, a substantial acceleration relative to the previous century. Table 3. Consumer Expenditure Surveys, 1909–1931. Health Share Medical Total 1909 2.7$780

$29,143 1914 2.6 869 33,619 1919 2.9 1,943 65,890 1921 3.1 1,849 59,031 1923 3.1 2,198 70,158 1925 3.0 2,396 79,303 1927 3.1 2,614 83,347 1929 3.2 2,904 89,370 1931 3.7 2,344 63,644 Sources: W. H. Lough (1935); Bureau of the Census (1976) Historical Statistics of the United States, tables G470, G486. The 1932 CCMC 28-volume report constituted a major advance in measurement methodology, yet line-item estimates made there show how different medicine was in 1929, often in ways that make comparisons with patterns of spending of today a stretch (see Table 4). Patients and families paid directly for more than three quarters of all expenditures, “physician services” being the largest category. Hospitals accounted for less than one quarter of spending, and most of that was for long-term care of mental and nervous disorders, disabilities, or tuberculosis rather than acute diseases. Retail sales of medical products, mostly tonics and elixirs that would today probably be banned by the FDA, amounted to one fifth of health spending. Prescriptions were informal notes that patients might take to a pharmacy as a suggestion for compounding but were not legally required. Insurance and prepayment were of great interest to the CCMC researchers, yet purchases were still so small they were lumped in with other hospital and physician fees.10 Medicine had begun the move toward complex, modern, organizational structures and financing, but only barely. Physician licensure was established in most states, but hospitals, pharmaceuticals, regulation, and insurance were still toddling out of infancy and had not yet taken on recognizably modern characteristics or expenditure patterns. Table 4: Committee on the Costs of Medical Care Estimates of Total Expenditures for Medical Care in 1929 TOTAL Patients Government Philanthropy Industry (Thousands) Physicians -$1,090,000

1,040,000

- - -

- - -

50,000

(private practice)

Dentists -

$445,000 445,000 - - - - - - - - - (private practice) Sectarian Practice$193,000

193,000

- - -

- - -

- - -

$142,000 142,000 - - - - - - - - - (private duty) Practical Nurses -$60,000

60,000

- - -

- - -

- - -

(private duty)

Hospitals -

$656,000 278,000 300,000 54,000 24,000 (operating expenses) Hospitals -$200,000

- - -

100,000

100,000

- - -

(new construction)

Public Health

$121,000 - - - 93,500 27,500 - - - Private Laboratories$3,000

3,000

- - -

- - -

- - -

Orthopedic Supplies

$2,000 2,000 - - - - - - - - - Glasses$50,000

50,000

- - -

- - -

- - -

Drugs

$665,000 665,000 - - - - - - - - - Organized Clinics -$29,000

7,790

16,000

210

5,000

(University, Industry, Army, Navy & Other)

TOTAL

$3,656,000 2,885,790 509,500 181,710 79,000 Source: Committee on the Costs of Medical Care (1932, table 5, p. 14). ## 1930–1955: Disruption and Consolidation Besides the Great Depression and World War II, the 25 years between 1930 and 1955 brought antibiotics, blood-banking, cardiac pacemakers, and other major medical advances, as well as substantial increases in the number and size of hospitals and a rise of employer-based health-insurance financing. What did not grow was the number of physicians relative to population. Medical-care expenditures per person rose, but the increase was more or less in line with rapidly rising wages. A study of expenditures by J. R. Seale (1959), published in The Lancet, covering the years 1929 to 1956 concluded: “The proportion of the gross national product of a nation devoted to medical care tends to remain constant. It rises during national economic depressions and it falls during wars. A persistent rise in real per-capita gross national product will tend to result in a very gradual increase in the proportion.” Depending on whether the starting point is taken to be 1930 or 1929, the estimated annual excess growth rate is 0.2% or 0.7% per year. Third-party payments, cross-subsidies and shadow pricing made estimates of national health expenditures increasingly problematic. Social Security Administration (SSA) reports indicate an excess growth rate increasing the health share of GDP by about 0.4% per year from 1930 to 1955 (Cooper et al., 1973). Census occupational data show healthcare rising from 1.8% to 2.5% of total employment from 1930 to 1950, an excess growth rate of more than +1.6% per year. Since this represented a major substitution of lower-cost ancillaries for physician labor, it is consistent with a much smaller rate of excess expenditure growth, perhaps well below +1.0% per year (Denison & Slater, 1943; Weinfeld, 1951). Supply was constrained by both the imposition of licensure and the limited capital for hospital expansion. George Stigler’s (1956) extensive National Bureau of Economic Research (NBER) study “Trends in Employment in the Service Industries” estimated 1950 health employment at 1.8 million, but it did not appear to note any excess growth in the medical labor force, and indirectly implied that growth in medical employment was no more rapid than in other service sectors—a sharp contrast to later BLS publications that termed healthcare a rapidly expanding “jobs machine” (Hiles, 1992; Kahl & Clark, 1986). The available sources each yield slightly different estimates of spending and growth rates during this span. Considered together, they appear consistent with a 1930 level of spending between 3.6% and 4.0% of GDP and annual excess growth rates averaging +0.1% to +0.9% per year. Despite the disruptions from the Great Depression and World War II, average personal incomes rose by 3.3% per year from 1930 to 1955, well above the long-run rate of 2.1% or the 1.3% annualized increase from 1910 to 1930; hence, large increases in medical expenditures could be made without increasing the health share of GDP (Field, 2003; Gordon, 2016).11 ## 1955–1975: Investments in Science, Hospitals, and Workforce Training Cause a Surge in Spending That Consolidates a National Health System ### Research Medical science and clinical practice clearly advanced during the first half of the 20th century, but more as a result of individual initiative and philanthropy than an organized research and development program. The National Institutes of Health (NIH) was founded in 1938, but had received only$464 million dollars in funding, a miniscule 0.0005% of GDP. By 1950, funding had grown 30-fold to a still rather small 0.017% of GDP (see Figure 2). The level was approximately the same in 1955; then it tripled to 0.074% in 1960, and doubled again in three years to 0.14% in 1963—after which funding grew at more or less the same rate as GDP for the next 30 years (Harden, 2015; Moses et al., 2015; Shryock, 1947b; Swain, 1962). Private research funding is more difficult to track, but it appears to have followed a similar trajectory. Pharmaceutical firms that in the early decades of the century had been mostly chemical companies compounding pills or purveyors of propriety elixirs and tonics shifted by the 1950s to become scientific laboratories relying on long-term research and the development of “ethical” drugs that required a prescription from a physician.

Figure 2. Funding of U.S. National Institutes of Health as % of GDP 1936–2016.

Sources: Swain (1962), Harden (2015), Moses et al. (2015).

### Hospitals

The effects of new technology are made apparent by a divergence of trends in the number of hospital beds and the cost-per-patient day (see Table 5).12 At the turn of the century, hospitals were often small additions connected to a physician’s residence or large state mental and infectious disease institutions. By mid-century, freestanding general acute-care facilities accounted for more than three quarters of hospital costs. Psychiatric and tuberculosis hospitals still had more beds and patients, but they received a much smaller share of resources.

Table 5: Hospitals: Beds, Days, Full-Time Equivalents, and Cost per Day, 1910–2015

Beds

% beds

FTE

days/

$cost % growth Hospitals per 1,000 General per bed person per day in 2009$

real cagr

1910

4359*

4.7

- - -

.4*

1930

6719*

7.8

41%

.7*

$5*$60

1946

6125

10.3

33%

1.1

.9

$9$76

2%

1950

6788

9.6

35%

1.4

.9

$14$102

7%

1955

6956

9.8

35%

1.5

.9

$21$135

12%

1965

7123

8.8

43%

1.9

1.1

$41$217

8%

1975

7156

6.9

65%

2.5

1.2

$133$424

12%

1985

6965

5.5

76%

3.1

1.0

$460$803

14%

1995

6291

4.1

81%

4.3

.8

$967$1,284

12%

2000

5810

3.5

84%

4.7

.7

$1,148$1,402

12%

2005

5756

3.2

85%

5.3

.7

$1,522$1,654

5%

2010

5754

3.0

85%

5.7

.6

$1,911$1,888

6%

2015

5627

2.8

87%

5.9

.6

$2,277$2,070

5%

Source: See note 12, in “Notes.”

The total number of beds per 1,000 population rose during the first five decades of the 20th century, then fell steadily after 1955. The American Hospital Association has since 1946 reported annually on the number of hospitals, beds, days of care, revenues, and expenses. At the time, the average cost per day was $9, or 0.5% of per-capita GDP. By 1955, costs had risen to$21 per day, or 0.8% of per-capita GDP. Ten years later, cost per day was $41, or 1.1% of per-capita GDP. By 1975, cost per day had risen to$133, or 1.7% of per-capita GDP. The number of full-time equivalents personnel per bed grew from 1.1 to 2.5. Tuberculosis hospitals disappeared, as did many state mental hospitals. The medical universe began to revolve around large academic medical centers that trained the most physicians, did most of the research, and were paid the most for each day of patient care. The era of high-tech, high-cost, inpatient therapeutics had arrived with force, becoming the main factor driving the health share of GDP upward. Costs per inpatient day continued to grow rapidly for the next 40 years, although the shift of surgery, rehabilitation, and palliative care toward outpatient settings have made “cost per day” less and less meaningful as an expenditure metric.

Hospitals, doctors, dentists, pharmaceutical companies, laboratories, imaging centers, and other entities supplying healthcare goods and services are counted as “users of funds” in the NHEAs and commonly referred to as “providers.”13 Allocations of government programs and insurance premiums are made to account for administrative and capital costs distinct from provider payments to avoid gaps or double-counting. Although physicians were the largest users of funds at the start of the century, hospitals have been the largest component since 1950. The bottom four rows of Table 6 are overhead and not directly attributable to individual patients. Administration and net cost of insurance has consumed a steadily larger portion over the last half-century, while the shares for investment spending on research and construction have been cut in half, falling from 3% to 1.5% and from 7% to 3.4%, respectively (see Table 6).

Table 6: Expenditure Shares by Provider Category, 1910–2015

Year

1910

1929

1955

1975

2000

2015

Physician

42%

30%

20%

19%

21%

20%

HOSPITAL

13%

18%

33%

38%

30%

32%

LTC

5%

6%

2%

9%

13%

13%

Rx Drugs

22%

18%

13%

6%

9%

10%

Dental & Prof

15%

17%

12%

7%

7%

6%

Other Med

4%

2%

9%

6%

4%

3%

-

?

3%

4%

6%

8%

Bldg & Equip

-

5%

4%

7%

4%

3%

Research

-

-

1%

3%

2%

1%

Public Health

-

3%

3%

2%

3%

3%

Sources: Centers for Medicare and Medicaid Services (2017a); Committee on the Costs of Medical Care (1932); Lough (1935); Reed & Hanft (1966), see also note 13 in “Notes.”

### Organization

The trend in medical practice over the last century has been toward ever-increasing organizational size and complexity (Cutler & Morton, 2013; Glied, Ma, & Solis-Roman, 2016; Moses et al., 2013). In 1900, the modal physician was a solo practitioner, receiving occasional support from a nurse or assistant. The relevance of solo practice dwindled as the practice of medicine became more complex. By 1970, there were two registered nurses and seven other health professionals, aides, and technicians for each doctor. Even though there were slightly fewer physicians per 1,000 population than in 1900, the proportion of total employment in the health sector had become four times as large (see Table 1).

Supervising health professionals required more training and more organizational capability. The doctor’s office was gradually replaced by the teaching hospital, which then morphed into the grander “academic medical center,” by 1970 (Dzau et al., 2010; Washington, Coye, & Feinberg, 2013). Research, teaching, community outreach, and advanced patient care were all brought together in one organization, though rarely under one roof. Capital expansion and mergers were so extensive that medical campuses stretched over many acres and were linked to satellite facilities miles away. Even though national health accounts have reported physician and hospital spending shares that were almost the same in 2016 as they were in 1960, management and administrative costs kept rising because provider size had increased dramatically.

Expenditure categories that served well from 1950 to 2000 are somewhat less useful for prior decades, or for the 21st century. The distinctions between physician and hospital, and between inpatient and outpatient, are becoming less meaningful. Transitions between home care, nursing homes, hospice, rehab, long-term acute care, and skilled nursing beds may be very important for reimbursement, but they are less so for understanding patient needs or therapy. Now that some states pay family members to assist disabled relatives, even the line between market and household production has again become porous. To understand and control future costs, it may be necessary to devise new accounting methods.

### Payment

Patients and families were the original payers of medical expenses, with limited supplementation from philanthropic and government funds. New forms of financial intermediation were required as costs rose from 3% of GDP to 6%. Health insurance was scalable, flexible, and robust—able to grow with the advancing technology and organizational complexity of medicine. More than half of the population had coverage by 1950, rising to 86% in 1970, when expansion stalled.14 Four decades later, 14% of the population was still uninsured (see Table 7). The Affordable Care Act of 2010 extended insurance to greater numbers of low-income persons who lacked employer plans and reduced the percentage of uninsured to less than 10%.

Table 7: Payer Financing, 1929–2015

1929

1950

1960

1970

1980

2000

2015

% with Insurance

< 3%

51%

72%

86%

89%

85%

89%

Personal Pay (OOP)

79%

56%

48%

33%

23%

15%

11%

Philanthropy etc.

6%

6%

7%

8%

8%

8%

7%

Government

14%

28%

24%

37%

42%

44%

49%

Medicare

10%

15%

16%

20%

Medicaid, CHIP

7%

10%

15%

17%

Private Insurance

1%

10%

21%

21%

27%

33%

33%

Sources: Centers for Medicare and Medicaid Services (2017a); Cohen et al. (2009); Committee on the Costs of Medical Care (1932); National Center for Health Statistics (2016); Numbers (1979); Reed and Hanft (1966); and note 14 in “Notes.”

The 20th-century transformation of medicine was accompanied by a transformation of the healthcare financing system. It is not so much that insurance “caused” health expenditures to grow as that a system of financing had to be built for health spending to grow, and once built, added momentum to rising consumer demand and technological advances (Peden & Freeland, 1998; Smith, Newhouse, & Freeland, 2009; Weisbrod, 1991). As third-party financing replaced personal payments, national health expenditures rose from $27 billion in 1960 to$75 billion in 1970. GDP per capita was growing rapidly, but health spending was growing even faster, with an annualized excess growth rate of +3.3%, reaching +5.2% during the peak years 1966 to 1970, amounts sufficient to have doubled the health share of GDP in 14 years had growth continued at that pace. Such growth was unprecedented and unmatched in any subsequent decade. The long period of postwar economic growth faltered in 1970 and then stumbled into stagflation and the OPEC (Organization of the Petroleum Exporting Countries) oil crisis of 1973. Health spending continued to rise, but it would never again reach the frenetic pace that occurred during the 1960s. After leveling out, growth became markedly slower toward the end of the century.

Public insurance becomes financially important after the passage of legislation establishing Medicare and Medicaid in 1965. Although private health insurance still covers a larger number of people, its funding role has declined because public insurance now provides more total funding, and Medicare has become the de facto standard for reimbursement, supplanting the Blue Cross and Major Medical methodologies that were more common from 1950 to 1980. Over half of all expenditures now come directly or indirectly from government sources. Twenty percent of “private” health insurance is for federal, state, and local employees, and the other 80% receive tax subsidies through IRS regulations. The CMS Office of the Actuary (CMS OACT) tabulations of funding by sponsor indicate that private business accounts for a declining share, now less than 20%, significantly under the 28% contributed by household personal payments and premiums.16

## After 1975: Bending the Cost Curve?

“Technology” is the short answer often given for why medical expenditures grew so rapidly. A more nuanced and verifiable answer is that a surge of investments in medical research, hospitals, and professional workforce training created new, high-technology academic medical centers that developed and disseminated advanced (and expensive) clinical practices, a process facilitated by the national expansion of third-party financing networks that vastly increased the flow of funds into the health system. The 1971 American Hospital Association manual Budgeting Procedures for Hospitals built on an assumption that expanded insurance coverage meant that most hospitals’ revenues could easily exceed any reasonable operating requirements (American Hospital Association, 1971). The manual even suggests that chief financial officers exercise restraint by setting charge levels well below what the market would bear, taking only the amount necessary to reach breakeven, plus a small surplus. This assumption of unlimited access to funding was being eroded even as the manual was being published. Economy-wide price controls were established under the Economic Stabilization Act of 1970, and special regulations for hospitals were issued in September 1973. This legislation represented the first of many national efforts at healthcare cost containment. As with most price controls, the immediate jolt failed to bring about the desired long-run results. The legislation was allowed to lapse in April 1974, before the planned Phase IV aggregate hospital cost controls could be implemented. The Health Maintenance Organization (HMO) Act of 1973 had a goal of bringing about cost control through lasting structural changes in the healthcare industry. Although the HMO Act did not live up to that promise, portions of the legislation and its many amendments continue to live on in current law. However inadequate and imperfect these cost-control rules and regulations were, their promulgation marked an end to the era of unfettered growth. Henceforth, budgets would always matter, even if legislated limits on spending were frequently breached (Mays 2007; U.S. Government Accountability Office, 1976).

The pace of growth moderated during the decades after 1970, averaging +2.8% per year from 1970 to 1990 and +1.8% per year from 1990 to 2010 (see Figure 3). Business-cycle fluctuations distort the measured health share of GDP, making it necessary to apply time series filters, smoothing, or distributed lag analysis to clarify the adjustment process. However, the spending restraint that kept measured health shares nearly constant at 13% of GDP from 1992 to 2000 does seem to be a result of real structural changes in healthcare rather than a reflection of macroeconomic fluctuations, since no major shocks distorted income or inflation over this period. Although the 1993 Clinton legislation proposal for healthcare reform was not enacted, the threat of regulation and the impact of the Balanced Budget Act of 1997 probably placed a check on expenditure growth. Even though a rebound after the turn of the century brought the health share above 15% by 2003, excess growth rates averaged less than +1% for the next four years, providing some evidence that a real reduction in trends had taken place. Assessing the slowdown in health spending since 2007 is more complicated. The relative importance of macroeconomic disruption and structural change has already been debated at length by economists and in the media.17 A headline that health spending grew “only” 4% in 2016 does not take account of the persistent decline in real incomes or the extraordinarily low rate of inflation. Disentangling these effects will require a more careful analysis of temporal dynamics and variable lags.

Figure 3: Population, real GDP, inflation and excess growth components of annual increase in U.S. National Health Expenditures 1960–2016.

# International Comparisons: OECD Data

Comparing patterns in the United States with other nations is beyond the scope of this review, but expenditure trends relative to major Organisation of Economic Co-operation and Development (OECD) countries are worth briefly noting. The discoveries and practices that initiated changes in the United States at the start of the 20th century came mostly from Europe, as did much of the population and economic organization.18 The innovations and organizational sophistication that accompanied the period of peak growth in national health expenditures during the 1960s, however, were generated internally. America had become the leader, instead of a follower, within a span of 50 years.

In a series of publications for the World Health Organization, Brian Abel-Smith (1967) observed that the United States spent roughly the same share of GDP on health as other developed countries in the early decades of the 20th century, then rose to 30% above the average by 1960 (4.7% vs. 3.6%). Expenditure shares in the United States remained 25% to 35% above the OECD median for the next 20 years. During the 1980s, the gap widened, exceeding 60% after 1990.19 The 2008 global recession strongly reduced expenditure trends in England, Germany, France, Japan, Australia, and Canada, as well as the United States (Morgan & Astolfi, 2013). As of 2015, the health share of GDP in the United States was 170% of the OECD median (17% vs. 10%). This divergence in national health spending between the United States and other developed countries is a major anomaly, one that has been addressed by many researchers but for which no consensus explanation has been reached. The lull that kept health expenditures relatively constant as a share of GDP from 1930 to 1955 in the United States appears to have occurred among OECD countries during that period as well. However, the 20 years of continued rapid growth after 1975 in the United States was not matched elsewhere. OECD median health shares expanded about +1% a year rather than the +2.5% in the United States, accounting for most of the accumulated 70% differential. A perception that extra spending has not brought commensurate extra value is spreading from researchers to policymakers and the public, suggesting that in future years, steps may be taken to reduce this disparity (Catillon, Cutler, & Getzen, 2018; Frakt, 2018).

# Temporal Dynamics and Variable Lags

Analysis of trends in health spending (H) over the long run relies on measuring changes in the health share (s) of GDP (Y) rather than nominal dollar amounts or real per-capita spending. Nominal comparisons across centuries are rendered almost meaningless by inflation and population increases. Even real per-capita cost estimates, so useful for cross-sectional comparisons or evaluation of growth over short time periods, become problematic over decades. Income levels so strongly affect both supply and demand that underlying shifts in spending patterns are overwhelmed and obscured. Focusing on health share concentrates attention on the main policy issue, excess growth, and avoids the complexity and errors of multiple adjustments for population, inflation, and income growth. Share data, tabulated by Carroll Wright in 1880, by the Committee on the Costs of Medical Care (CCMC) in 1929, and by the Bureau of Economic Analysis (BEA) and the Centers for Medicare and Medicaid Services Office of the Actuary (CMS OACT) in 1960 or 2016, all reflect the relative importance of healthcare in consumer and government budgets.

$Display mathematics$

Unfortunately, volatility in the denominator (GDP) tends to add noise, distorting the measured excess growth rate (% $Health%GDP). The healthcare system is inertial, responding only slowly to macroeconomic business cycles, with an average lag of three to six years, as is seen in Figure 4(a) & 4(b) and Table 8.20 Table 8: Annual Health Expenditure Growth Regression on Lagged Real Per-Capita GDP and Inflation, 1960–2015 Constant real per-capita GDP growth. Deflator Time R2 year 0 year −1 year −2 year −3 year −4 year −5 year 0 year -1 % growth NHE .046 .17 .07 .04 .19 .29 .23 .28 .12 −.0006 .702 Sources: Getzen (1990, 2014, 2016a, 2016b), and note 20 in “Notes.” Figure 4. Annual growth of U.S. National Health Expenditures relative to (a) Current income (real per capita GDP growth); (b) Lag smoothed income (real GDP per capita = 6-year trailing MA, Inflation = 3-year trailing MA). Smoothing with a six-year moving average of real per-capita income and a three-year weighted average of inflation removes cyclical and random noise, making the underlying trends more apparent, and provides a much better fit (r2 = .616) than using contemporaneous income and inflation (r2 = .113; Getzen, 1990, 2014, 2016a; Smith, Freeland, Heffler, McKusick, & Health Expenditures Projection Team, 1998). The arrival of the Great Recession 2008–2010 made it abundantly clear that business cycles affect national health spending. In earlier decades, multiyear lags made the delayed and gradual response to macroeconomic disruption by the health sector less visible. The possibility that structural rigidities in the health professions and reimbursement create lags lasting much longer than the standard business cycle should be considered. The Great Depression and other factors restrained growth for several decades after 1929, and then the upsurge in medical facilities, discoveries, and workforce continued to push growth upward for several decades after 1968. # Population Aging and the Allocation of NHE In 1900, just 4% of the U.S. population were age 65 or older (65+), a share that doubled to 8% by 1950 and reached 12.5% in 2000. The offsetting demographic effects from the Depression-era birth dearth and the postwar baby boom kept the fraction of 65+ relatively constant from 1990 to 2010. Since then, it has risen steadily, and it is expected to reach 16% by 2020 and to move above 20% by 2050 before stabilizing over the long run. Even though the elderly were more apt to need medical care, the limited ability of families to pay for such care, and the relatively small share of the population to reach advanced ages meant that only a small part of total health spending during the first half of the twentieth century was attributable to older Americans. The earliest survey with reliable age-related spending estimates was conducted in 1953. Average spending was$110 over age 65 and $67 under age 65, a ratio of 1.7:1 (see Figure 5 and Table 9).21 At mid-century, 8.5% of the total U.S. population and 13% of estimated national personal-health spending was accounted for by the older age group. In the year 2000, 12.5% of the population was age 65+; the old:young spending ratio was 4.4:1; and the share of spending for older patients was 39%, three times as large as it had been a half-century earlier. During the next decade, the ratio of spending for the elderly relative to the young declined to 3.3:1, reducing the expenditure share of the elderly to 34% even as population continued to age. Table 9: Personal Healthcare Spending Per Capita by Age, Selected Years 1953–2012 ratio % pop % spending age 0-64 age 65+ 65+/0-64 age 65+$$age 65+ 1953$67

$110 1.7 8.5% 13% 1963 129 304 2.4 9.4% 20% 1965 158 472 3.0 9.5% 24% 1967 171 528 3.1 9.8% 25% 1970 238 823 3.5 10.8% 30% 1977 453 2,002 4.4 12.2% 38% 1987 1,088 5,849 5.4 12.7% 44% 1996 2,123 10,308 4.9 12.4% 41% 2000 2,676 11,815 4.4 12.5% 39% 2002 3,521 13,537 3.8 12.4% 35% 2004 4,062 15,112 3.7 12.4% 34% 2006 4,577 16,434 3.6 12.5% 34% 2008 4,998 17,786 3.6 12.8% 34% 2010 5,381 18,544 3.4 13.1% 34% 2012 5,781 18,988 3.3 13.5% 34% Sources: See note 21 in “Notes.” Figure 5: Ratio of personal healthcare spending ages 65+ to young persons age 0–64. Sources: Cutler and Meara (1997), Meara, White, and Cutler (2004), Lassman et al. (2014), CMS (2017b). The share of GDP spent on medical care for the older population rises as the percentage age 65+ increases, as average per-capita medical expenditures rise, or as spending per person on the elderly rises relative to the young. Macroeconomic growth accounting decomposition is extended in Table 10 to examine these effects. National health expenditures rose from$35 billion in 1963 to $2,795 billion in 2012, an average annual growth rate of 9.4%, attributable to a 6.8% GDP growth rate (1% population, 2% real income per capita, and 3.7% inflation) plus an excess growth rate of +2.4% for the health sector. Only 0.3% of the overall increase is attributable to the elderly; +0.15% is due to the rise in the fraction of the population over age 65; and another +0.15% per year is due to the relatively more rapid increase in spending per person over age 65 (the ratio rises from 2.4:1 to 3.3:1).22 Table 10: Increase in NHE Decomposed by Age Group, 1963–2012 1963–2012 63–87 87–12 Annual % Growth of NHE 9.4% 11.9% 7.0% Population 1.0% 1.1% 1.0% Inflation 3.7% 5.1% 2.3% real per-capita GDP 2.0% 2.5% 1.5% Excess (Share growth) 2.4% 2.8% 2.0% Contribution of population aging +0.3% +1.3% (-0.6%) Aging Effect Decomposition average$ growth 0-64

2.1%

1.5%

2.6%

change in % population 65+

0.15%

0.3%

0.1%

change in price ratio

0.15%

1.0%

-0.7%

Sources. See Table 9 and notes 21 and 22).

In 1987, the elderly were 12% of the total population but accounted for 44% of total health spending. That share slowly declined for the next 20 years and then stabilized, but it is likely to rise again in coming decades and to exceed 50% after 2050. The rise and fall of age-related health spending has not always been fully appreciated by analysts or the public. Since both the rate of growth in the share of population age 65+ and the ratio of per-capita spending by the age 65+ group relative to the rest of the population has varied over time, the estimated contribution of population aging depends on the period chosen for analysis. During the span from 1963 to 1987, two thirds of the increase in the health share of GDP, from 5.4% to 10.6%, was attributable to the over 65 group, twice as much as all the spending by patients aged 0–64 combined (+3.4% vs. +1.7%).23 After 1987, the older population actually reduced the average trend in per-capita spending. Even though age 65+ population rose from 12.2% to 13.5%, generating a rise in spending of 0.1% per year because older persons cost more, this demographic effect was more than offset by a relative cost decline. Expenditures per person 0–64 were rising so much faster that the old:young ratio of costs fell from 5.4x to 3.3x, reducing the trend by −0.7% per year. An “age wave” of costly healthcare for elders swelled and then passed. The rise and fall were primarily a result of policy, caused by changes in reimbursement, not a result of demographic changes or the incidence of disease.

Given that most of the rapid rise in medical costs during the 1970s and 1980s was attributable to spending for patients over age 65, it is not surprising that so many analysts intoned “demography is destiny” and attempted to explain rising expenditures as a function of population aging per se. Morris Barer and Robert Evans (Barer et al., 1989) were among the first to point out that it was increased spending per elderly person, not aging populations, that was actually causing expenditures to rise so fast. Subsequently, Getzen (1992) used OECD data to demonstrate that neither the level nor the rate of growth in percentage of population age 65+ was significantly correlated with international differences in per-capita health-spending care once the income effects had been controlled for. More and more researchers reported minimal aging effects, as observations were extended into the 1990s, when the share of total spending attributable to the over 65 group had begun to decline (Chernichovsky & Markowitz, 2004; Reinhardt, 2003). Although the argument that aging is a major determination of national health spending is still sometimes heard, previous views regarding its importance have faded. However, the fact that retirees, rather than workers and their children, will eventually account for more than half of all patient expenditures still poses a major social and political conundrum.

# Projections to 2026 and Beyond

Computationally and conceptually, long-run projections of expenditures 10 years or more into the future take the form of GDP + X%. The forecasts made by health economists and actuaries are conditional; given the available population and macroeconomic projections and assuming that medical technology continues to advance, then excess growth of GDP + X% per year is projected. The Centers for Medicare and Medicaid (CMS) Office of the Actuary, the Congressional Budget Office (CBO), and the Society of Actuaries (SOA) routinely prepare and publish long-run NHE projections, and all three have similar structure: a 10-year estimated share of GDP, an intermediate transition (termed a “step-down” period by actuaries), and an eventual convergence to GDP + 0%, since any other asymptote would make the model unstable by implying that the health sector would eventually consume the entire economy (or disappear).24 Nominal spending amounts are heavily affected by macroeconomic inflation and income assumptions. The Congressional Budget Office, Centers for Medicare and Medicaid Services, and Society of Actuaries projections make use of the same sources, routinely documented in the Medicare Trustees Report, facilitating comparisons of estimated excess growth rates. In contrast to the standardized structure and comparability of long-run projections, short-term forecasts of spending for the next one to five years can differ widely in methodology and detail, and often rely heavily on professional judgment or particular insurance benefit stipulations and reimbursement factors. CMS and CBO estimates are mandated by Congress to be “current law” projections and of necessity must focus on the Medicare and Medicaid populations and programs funded by the federal government. Such specific factors, restrictions, and judgmental adjustments are very important for short-run estimates but meld into the overall average spending trend within a few decades. Beyond a ten-year horizon, what matters is the expected future health share of GDP, which is a function of the cumulative annualized excess growth rate.

Ten-year excess growth rates have averaged +0.9% to +2.7% for the last 25 years, leading to expectations that the trend will continue within that range for the next ten years. Growth has slowed somewhat during more recent decades, albeit erratically, making it more likely that future increases will be toward the lower end of the range. CMS has projected ten-year annualized forward excess growth rates of about 1.1% ±0.2% since 2009, the most recent being 1.14% for 2015 to 2025 (Centers for Medicare and Medicaid Services, 2017a). Even if major structural changes in the health system were to begin one year after this data was analyzed, or had already started a few years ago, measures of expenditure growth from now until 2020 or 2026 might not reveal such a tectonic shift. A change from +1.2% down to +0.6% or up to +1.8% might not be readily distinguishable from ordinary variation created by macroeconomic disturbances and random noise. Relative errors in the 2026 share projections are likely to be smaller than the errors in forecasting nominal or real GDP, or post hoc revisions to the present 2016 NHE estimate.25 Given the budgetary outlook for 2020, the health share will probably fall somewhere between 18% and 23% of GDP, which would be consistent with either random fluctuations around existing trends or the early part of a new regime.

At some point in the future, another major system change will occur. It has been more than 50 years since the last transformation in the 1960s. Given the lack of public satisfaction with the U.S. healthcare system expressed in 2018, it could be argued that restructuring is likely to come sooner rather than later. The next system change may or may not coincide with a distinctive change in medical technology, but it will very likely be marked by and include a change in financing mechanisms and information technology. Average spending levels have risen from 5% of income in the 1960s toward 20% by 2020, making private health insurance much less affordable to working families. Income growth appears to be declining and concentrated toward the top. It is no longer just the poor who require subsidies; it is also the median household with less than $60,000 to support multiple family members. Marginal productivity of medical expenditure appears to be falling. Every additional year of life expectancy gained seems to be more and more costly (Cutler et al., 2006). Lifestyle, environment, exercise, inequality, family relationships, and other social determinants are increasingly seen as more important for health than additional spending on clinical therapeutics. The design of health financing in 1965 focused on employer-provided health insurance, with safety-net add-ons for the poor, the elderly, and the disabled. In 2020, those add-ons will account for more than half of all expenditures. The demise of defined-benefit employer pensions over the last two decades suggests weaker, not stronger, financial support from corporations ( Butrica, Iams, Smith, & Toder, 2009; Jaffe, 2004). Previous reliance on voluntary private insurance is probably not sustainable. Taken together, these changes in conditions and constraints over the last five decades would appear to make a further deceleration in medical expenditure growth more likely than a sudden acceleration. # Conclusion: Trend Shifts and System Change The growth of healthcare in the United States divides readily into three major phases: a long fallow period, decades of transition, and then a modern era expanding within a structure that is more or less stable. Medicine was an enduring but therapeutically ineffective and economically minor presence for centuries. Employment in health occupations was relatively steady at 0.8% of total U.S. employment during the first 30 years it was measured, 1850–1880, and then began to rise in the next decade—a quantified signal of growth. The turning point from prescientific clinical nihilism toward early technological advances can be placed at around 1910. The marker could arguably be put a decade or two before or after since the process of change was gradual and uneven, and since the difference between decades was less than the difference between urban and rural practices during this era. The transition to a modern healthcare system depended on three developments: (a) national standards for medical practice, (b) institutional networks linking medical education and organization, and (c) broad financial support with national subsidies. The Flexner Report called for national standards of medical training, including college chemistry, biology, and physics, as well as a hospital internship. During the 1920s and 1930s, these educational requirements were incorporated in state licensure restrictions. Idiosyncratic judgments by solo physicians were slowly replaced by scientific therapies, but the definition of what constituted “appropriate” medical care still relied on the judgments of local county medical societies. Mobilization for war in the 1940s regularized and routinized medical practice across the nation, helping to form a deep administrative structure. Provider organizations and health insurance replaced doctors working autonomously and receiving payment from individual patients. Hospitals and academic medical systems became powerful players, defining what constituted good medical care. Yet ultimately, it was the power of the purse that fused a truly national healthcare system. IRS regulations subsidized the growth of not-for-profit hospitals and private-employer insurance plans in the 1940s and 1950s (Thomasson, 2002). Legislators provided the capital necessary for major expansion and took responsibility for financing a safety net, ensuring care for the elderly, poor, and disabled. Without Medicare and Medicaid certification (and funding) most hospitals and nursing homes could no longer survive. After 1965, segregated and substandard facilities were pushed into compliance as control over payment provided the power to enforce standards and shape medical organization (Smith, 2015). Although it was a patchwork system, cobbled together from a path-dependent mix of public and private parts, with significant tension between state and federal control, healthcare in the United States had become undeniably national. Culmination of the transition phase can be placed within the span of a few years. Growth peaked in 1968–1970 following a surge of expenditures powered by investments in capital structures, workforce training, and research. The passage of Medicare and Medicaid legislation led rather quickly to “almost” universal insurance coverage, supplementing the private health-insurance plans developed over the preceding decades. Access to advanced medical treatments became part of the social fabric—a much denser and richer social fabric now that most Americans had automobiles and either lived near a hospital or could easily drive to one. Life expectancy rose above the traditional three-score-and-ten. As the pace of economic growth slowed in the following decades, calls to control medical costs accumulated. Although the healthcare system continued to expand, it remained recognizably similar for the next 45 years—highly organized and regulated, with academic medical centers at the core and “almost” universal insurance coverage for more than 80% of the population, who subsidized a safety net for the remainder. Marking major regime changes at 1910 and 1970, Table 11 also indicates intermediate secondary periods when expenditure trends appear to have shifted. Table 11: Eras in the Growth of U.S. Health Expenditures I. Prescientific II. Transitional III. Postmodern a) early b) late a) early b) formative c) surge a) early b) present 1776–1880 1910 1929 1955 1970 1995 2015 H$/GDP

.026

.029

.035

.042

.069

.133

.178

growth rate

0.4%

1.0%

0.7%

3.5%

2.7%

1.4%

Since 1995, excess growth rates have been cut in half, although it is hard to point to specific technological, organizational, or policy changes that can explain such significant deceleration, nor is there an expert consensus on why the slowdown occurred. The 45 years from 1970 to 2015 could be viewed as a single period during which health spending gradually, albeit erratically, wound down while the institutional structures and financing mechanisms stayed mostly the same, or divided into two periods as is done in Table 11. Linkage between resource expenditures and regime change appears to be either very loose or subject to long lags. The macro-generalizations, trends, and stylized facts reiterated in this concluding section are robust, and sufficiently supported by empirical evidence that any explanation or theory of long-run health-spending trends should be consistent with them.

Anderson, O. W. (1990). Health services as a growth enterprise in the United States since 1875. Ann Arbor, MI: Health Administration Press.Find this resource:

Carter, S. B., Gartner, S. S., Haines, M. R., Olmstead, A. L., Sutch, R. & Wright G. (Eds.). (2006). Historical Statistics of the United States: Millennial Edition. New York, NY: Cambridge University Press.Find this resource:

Catillon, M., Cutler, D., & Getzen, T. (2018). Two hundred years of health and medical care: The importance of medical care for life expectancy gains (Working Paper 25330). National Bureau of Economic Research. Cambridge, MA.Find this resource:

Catlin, A. C., & Cowan, C. A. (2015). History of health spending in the United States, 1960–2013. Centers for Medicare and Medicaid Services, Baltimore, MD.Find this resource:

Centers for Medicare and Medicaid Services (CMS). National Health Expenditure Data. Baltimore, MD.Find this resource:

Chandra, A., & Skinner, J. (2012). Technology growth and expenditure growth in health care. Journal of Economic Literature, Medicine and Society in America, 50, 645–680.Find this resource:

Costa, D. (2015). Health and the economy in the United States from 1750 to the present. Journal of Economic Literature, 53, 503–570.Find this resource:

Getzen, T., & Okunade, A. (2018). Symposium introduction: Papers on “modeling national health expenditures.” Health Economics, 26, 827–833.Find this resource:

Starr, P. (2008). Social transformation of American medicine: The rise of a sovereign profession and the making of a vast industry. New York, NY: Basic.Find this resource:

Stevens, R. (1998). In sickness and in wealth: American hospitals in the twentieth century. Baltimore, MD: Johns Hopkins University Press.Find this resource:

Vogel, M., & Rosenberg, M. (1979). The therapeutic revolution: Essays in the social history of American medicine. Philadelphia, PA: University of Pennsylvania Press.Find this resource:

## References

Abel-Smith, B. (1967). An international study of health expenditure (Public Health Paper). Geneva, Switzerland: World Health Organization.Find this resource:

American Hospital Association. (1971). Budgeting procedures for hospitals. Chicago, IL: American Hospital Association.Find this resource:

Anderson, O. W. (1963). Changes in Family Medical Care Expenditures and Voluntary Health Insurance: A Five-Year Resurvey. Cambridge, MA: Harvard University Press.Find this resource:

Anderson, O. W. (1990). Health services as a growth enterprise in the United States since 1875. Ann Arbor, MI: Health Administration Press.Find this resource:

Anderson, O. W., Sinai, N., & Dollar, M. L. (1946). Health insurance in the United States. New York, NY: Commonwealth Fund.Find this resource:

Arrow, K. J. (1963). Uncertainty and the welfare economics of medical care. American Economic Review, 53, 141–149.Find this resource:

Barer, M., Pulchins, I. R., Evans, R. G., Hertzman, C., Lomas, J., &.Anderson, G. M. (1989). Trends in the use of medical services by the elderly in British Columbia. Canadian Medical Association Journal, 141(1), 39–45.Find this resource:

Beck, A. J. (2004). The Flexner Report and the standardization of American medical education. JAMA, 291, 2139–2140.Find this resource:

Bernard, D., Cowan, C., Selden, T., Cai, L., Catlin, A., & Heffler, S. (2012). Reconciling medical expenditure estimates from the MEPS and NHEA, 2007. Medicare and Medicaid Research Review, 2(4), E1–E20.Find this resource:

Blumenthal, D., Stremikis, K., & Cutler, D. (2013). Health care spending: A giant slain or sleeping? New England Journal of Medicine, 396, 2551–2557.Find this resource:

Bryan, L. S. (1964). Blood-letting in American medicine, 1830–1892. Bulletin of the History of Medicine, 38, 516–529.Find this resource:

Bureau of the Census. (1975). Historical Statistics of the United States Colonial Times to 1970. Washington, DC: U.S. Government Printing Office.Find this resource:

Bureau of the Census. (2018). 2010 Census Urban and Rural Classfication and Urban Area Criteria.Find this resource:

Bureau of Labor Statistics. (1920). Standards of living: A compilation of budgetary studies. Bureau of Applied Economics, Washington, DC.Find this resource:

Bureau of Labor Statistics. (1924, May). Cost of living in the United States. Bulletin of the United States Bureau of Labor Statistics, No. 357. Washington, DC: U.S. Government Printing Office.Find this resource:

Bureau of the Labor Statistics. (2003). Handbook of U.S. Labor Statistics. Washington, DC: U.S. Government Printing Office.Find this resource:

Bureau of Economic Analysis. (2017). NIPA Handbook: Concepts and methods of the U.S. National Income and Product Accounts.Find this resource:

Butrica, B. A., Iams, H. M., Smith, K. E., & Toder, E. J. (2009). The disappearing defined benefit pension and its potential impact on the retirement income of baby boomers. Social Security Bulletin, 69(3), 1–27.Find this resource:

Bynum, W. F. (1994). Science and the practice of medicine in the nineteenth century. New York, NY: Cambridge University Press.Find this resource:

Carter, S. B., Gartner, S. S., Haines, M. R., Olmstead, A. L., Sutch, R. & Wright G. (Eds.). (2006). Historical statistics of the United States, millennial edition. New York, NY: Cambridge University Press.Find this resource:

Catillon, M., Cutler, D., & Getzen, T. (2018). Two hundred years of health and medical care: The importance of medical care for life expectancy gains. (NBER Working paper No. 25330). National Bureau of Economic Research. Cambridge, MA.Find this resource:

Catlin, A. C., & Cowan, C. A. (2015). History of health spending in the United States, 1960–2013. Centers for Medicare and Medicaid Services, Baltimore, MD.Find this resource:

Centers for Medicare and Medicaid Services. (2010). 2010 Comprehensive Revision. Office of the Actuary, Baltimore, MD.Find this resource:

Centers for Medicare and Medicaid Services. (2015). National health expenditure accounts: Methodology paper, 2015; definitions, sources and methods.Find this resource:

Centers for Medicare and Medicaid Services. (2017a). National Health Expenditure Data. Centers for Medicare and Medicaid Services, Baltimore, MD.Find this resource:

Centers for Medicare and Medicaid Services. (2017b). Projections of national health expenditures: Methodology and model specification. Office of the Actuary, Baltimore, MD.Find this resource:

Centers for Medicare and Medicaid Services. (2017c). Age and gender tables. National Health Expenditure Age and Gender data.Find this resource:

Congressional Budget Office. (2017). The 2017 long-term budget outlook. Washington, DC: Congressional Budget Office.Find this resource:

Chandra, A., Holmes, J., & Skinner, J. (2013). Is this time different? The slowdown in healthcare spending. (NBER Working paper 19700). National Bureau of Economic Research. Cambridge, MA.Find this resource:

Chandra, A., & Skinner, J. (2012). Technology growth and expenditure growth in health care. Journal of Economic Literature, 50, 645–680.Find this resource:

Chen, A., & Goldman, D. (2016). Health care spending: Historical trends and new directions. Annual Review of Economics, 8(1), 291–316.Find this resource:

Chernew, M. (2014, September 3). Examining the present and future of the health spending growth slowdown. Health Affairs Blog.Find this resource:

Chernew, M., & Newhouse, J. (2012). Health care spending growth. In M. Pauly, T. G. McGuire, & P. P. Barros (Eds.), Handbook of health economics (Vol. 2, pp. 1–43). Amsterdam, The Netherlands: North Holland.Find this resource:

Chernichovsky, D., & Markowitz, S. (2004). Aging and aggregate costs of medical care: Conceptual and policy issues. Health Economics, 13, 543–562.Find this resource:

Cochrane, J. H. (1994). Shocks. Paper presented at the Carnegie-Rochester Conference Series on Public Policy, 41, 295–364. North Holland: Elsevier.Find this resource:

Cochrane, J. H. (2016). Macro-finance (NBER Working Paper 22485). National Bureau of Economic Research. Cambridge, MA.Find this resource:

Cohen, R., Makuc, D. M., Bernstein, A. B., Bilheimer, L. T., & Powell-Griner, E. (2009). Health insurance coverage trends, 1959–2007 (National Health Statistics Report No. 17). Hyattsville, MD: National Center for Health Statistics.Find this resource:

Coile, D. (2014). GDP: A brief but affectionate history. Princeton, NJ: Princeton University Press.Find this resource:

Committee on the Costs of Medical Care. (1932). Medicine and society in America: Final report of the CCMC. Chicago, IL: University of Chicago Press. This report is also available in 1972 Arno Press edition.Find this resource:

Congressional Budget Office. (2017). The 2017 long-term budget outlook. Washington, DC: Congressional Budget Office.Find this resource:

Consoli, D., & Mina, A. (2009). An evolutionary perspective on health innovation systems. Journal of Evolutionary Economics, 19, 297–319.Find this resource:

Cooper, B., Worthington, N. L., & McGee, M. F. (1973). Compendium of national health expenditures data (DHEW No. (SSA)73-11903). Office of Research and Statistics. Washington, DC: U.S. Government Printing Office.Find this resource:

Cunningham, R. C., III, & Cunningham, R. M. (1997). The blues: A history of the Blue Cross and Blue Shield system. DeKalb: Northern Illinois University Press.Find this resource:

Cutler, D., & Meara, E. (1997). Medical costs of the young and old: A forty-year perspective. (NBER Working paper No. 6114). National Bureau of Economic Research. Cambridge, MA.Find this resource:

Cutler, D., & Morton, F. S. (2013). Hospitals, market share, and consolidation. JAMA, 310, 1964–1970.Find this resource:

Cutler, D., Rosen, A. B., & Vijan, S. (2006). The value of medical spending in the United States, 1960–2000. New England Journal of Medicine, 355, 920–927.Find this resource:

Cutler, D., & Sahni, N. (2013). If slow rate of health care spending growth persists, projections may be off by $770 billion. Health Affairs, 32, 841–850.Find this resource: Deaton, A. (1977). Involuntary saving through unanticipated inflation. American Economic Review, 67, 899–910.Find this resource: Deaton, A. (2016). Measuring and understanding behavior, welfare, and poverty (Nobel lecture). American Economic Review, 106, 1221–1243.Find this resource: Denison, E. F., & Slater, A. (1943, October). Incomes in selected professions: Part 4, medical service. Survey of Current Business, 23(10), 16–20.Find this resource: Dickinson, F. G. (1951). Medical care expenditures, prices and quantity, 1930–1950. JAMA, 147, 1354–1359.Find this resource: Dickinson, F. G., & Raymond, J. (1955). The economic position of medical care 1929–1953. MA, 159(1), 41–50, and 159(2), 119–127.Find this resource: Dranove, D. (2009). The economic evolution of American health care. Princeton, NJ: Princeton University Press.Find this resource: Dublin, L. I. (1927, November 1). What price doctors? Harper’s Magazine, 767–776.Find this resource: Dzau, V. J., Ackerly, D. C., Sutton-Wallace, P., Merson, M. H., Williams, R. S., Krishnan, K. R., Taber, R. C., & Califf, R. M. (2010). The role of academic health science systems in the transformation of medicine. The Lancet, 375, 949–953.Find this resource: Edwards, A. M. (1943). Comparative occupational statistics for the United States, 1870 to 1940. Report U.S. Bureau of the Census. Washington, DC: U.S. Government Printing Office.Find this resource: Fetter, B. (2006). Origins and elaboration of the national health accounts, 1926–2006. Health Care Financing Review, 28(1), 53–67.Find this resource: Field, A. (2003). The most technologically progressive decade of the century. American Economic Review, 93, 1399–1411.Find this resource: Fisher, C. C. (1980). Difference by age groups in health care spending. Health Care Financing Review, 1(Spring), 65–90.Find this resource: Flexner, A. (1910). Medical education in the United States and Canada. Report to the Carnegie Foundation, New York, NY.Find this resource: Fogel, R. W., Fogel, E. M., Guglielmo, M., & Grotte, N. (2013). Political arithmetic: Simon Kuznets and the empirical tradition in economics. Chicago, IL: University of Chicago Press.Find this resource: Frakt, A. (2018, May 14). Medical mystery: Something happened to U.S. health spending after 1980. New York Times.Find this resource: Frandsen, B., Powell, M., & Rebitzer, J. B. (2017). Sticking points: Common-agency problems and contracting in the U.S. healthcare systems (NBER working paper No. 23177, 42 pages). National Bureau of Economic Research. Cambridge, MA.Find this resource: Garbarino, J. W. (1959). Price behavior and productivity in the medical market. Monthly Labor Review, 13(1), 3–15.Find this resource: Garber, A., & Skinner, J. (2008). Is American health care uniquely inefficient? Journal of Economic Perspectives, 22(4), 27–50.Find this resource: Gerdtham, U., & Jönsson, B. (2000). International comparisons of health expenditure: Theory, data and econometric analysis. In T. Culyer & J. Newhouse (Eds.), Handbook of health economics (pp. 12–53). Amsterdam, The Netherlands: Elsevier Science.Find this resource: Gerdtham, U., Sogaard, J., Andersson, F., & Jonsson, B. (1992). An econometric analysis of health care expenditure: A cross-section study of the OECD countries. Journal of Health Economics, 11(1), 63–84.Find this resource: Getzen, T. E. (1985). Reducing healthcare costs. Journal of Healthcare Financial Management, 39(3), 34–38.Find this resource: Getzen, T. E. (1990). Macro forecasting of national health expenditures. In L. Rossiter & R. Scheffler (Eds.), Advances in health economics (Vol. 11, pp. 27–48). Greenwood, CT: JAI Press.Find this resource: Getzen, T. E. (1992). Population aging and the growth of health expenditures. Journal of Gerontology: Social Sciences, 47(3), S98–104.Find this resource: Getzen, T. E. (2000). Health care is an individual necessity and a national luxury: Applying multilevel decision models to the analysis of health care expenditures. Journal of Health Economics, 19(2), 259–270.Find this resource: Getzen, T. E. (2014). Macroeconomic dynamics of health: Lags and variability in mortality, employment and spending. In A. J. Culyer (Ed.), Encyclopedia of health economics (Vol. 2, pp. 165–176). San Diego, CA: Elsevier.Find this resource: Getzen, T. E. (2016a). Measuring and forecasting global health expenditures. In R. Scheffler (Ed.), Global handbook of health economics (Vol. 1, pp. 177–215). London, U.K.: World Scientific.Find this resource: Getzen, T. E. (2016b). Accuracy and range of actuarial projections of health care costs. North American Actuarial Journal, 20(2), 101–113.Find this resource: Getzen, T. E. (2017). Health spending in the USA: 1776 to 2026. SSRN.Find this resource: Glied, S., Ma, S., & Solis-Roman, C. (2016). Where the money goes: Evolving expenses of the US health care system. Health Affairs, 35, 1197–1203.Find this resource: Gordon, R. J. (2016). The rise and fall of American growth: The U.S. standard of living since the Civil War. Princeton, NJ: Princeton University Press.Find this resource: Haber, S. G., & Newhouse, J. P. (1991). Recent revisions to and recommendations for national health expenditures accounting. Health Care Financing Review, 13(1), 111–116.Find this resource: Hacker, J. D. (2010). Decennial life tables for the white population of the United States, 1790–1900. Historical Methods, 43, 45–79.Find this resource: Haines, M. R. (2006). Vital Statistics. Pp. 1–381 to 390 in Carter et al (2006).Find this resource: Haines, M. R., Steckel, R. (2000). A population history of North America. Cambridge, U.K.: Cambridge University Press.Find this resource: Hall, R. E., & Jones, C. I. (2007). The value of life and the rise in health spending. Quarterly Journal of Economics, 122(1), 39–72.Find this resource: Harden, V. A. (2015). A short history of the National Institutes of Health. Bethesda, MD: National Institutes of Health.Find this resource: Hartman, M., Catlin, A., Lassman, D., Cylus, J., & Heffler, S. (2008). U.S. health spending by age: Selected years through 2004. Health Affairs, 27, w1–w12.Find this resource: Hartman, M. B., Kornfeld, R. J., & Catlin, A. C. (2010). A reconciliation of health care expenditures in the national health expenditures accounts and in gross domestic product. Survey of Current Business, 90(9), 42–52.Find this resource: Hiles, D. R. H. (1992, November). “Health services: The real jobs machine.” Monthly Labor Review, 3–16.Find this resource: Horenstein, A. R., & Santos, M. (2016). Understanding growth patterns in US health care expenditures. SRRN.Find this resource: Jaffe, J. (2004) The decline of private-sector defined benefit promises and annuity payments: What will it mean? EBRI Notes, 25(7), 2–8.Find this resource: Kahl, A., & Clark, D. (1986, Aug). Employment in health services: Long-term trends and projections. Monthly Labor Review, 17–36.Find this resource: Keehan, S. P., Lazenby, H. C., Zezza, M. A., & Catlin, A. C. (2004). Age estimates in the National Health accounts.” Health Care Financing Review, 1(1), 1–16.Find this resource: Keehan, S. P., Stone, D. A., Poisal, J. A., Cuckler, G. A., Sisko, A. M., Smith, S. D. . . . , Lizonitz, J. M. et al. (2017). National health expenditure projections, 2016–25: Price increases, aging, push sector to 20 percent of economy. Health Affairs, 36, 1–11.Find this resource: Kendix, M., & Getzen, T. E. (1994). U.S. health services employment: A time series analysis. Health Economics, 3(3), 169–181.Find this resource: Kett, J. F. (1968). The formation of the American medical profession: The role of institutions, 1780–1860. New Haven, CT: Yale University Press.Find this resource: King, L. S. (1991). Transformations in American medicine: From Benjamin Rush to William Osler. Baltimore, MD: Johns Hopkins University Press.Find this resource: Klarman, H. E. (1977). The financing of health care. Daedalus, 106(1), 215–234.Find this resource: Klem, M. C. (1942, August). Some recent developments in voluntary health insurance. Social Security Bulletin, 5–16.Find this resource: Langford, E. A. (1957). Medical care in the Consumer Price Index, 1936–56. Monthly Labor Review, 80, 1053–1058.Find this resource: Lassman, D., Hartman, M., Washington, B., Andrews, K., & Catlin, A. (2014). Health spending trends by age and gender: Selected years 2002–2010. Health Affairs, 33, 815–822.Find this resource: Levitt, L., Claxton, G., Roehrig, C., Getzen, T. (2013, April 22). Assessing the effects of the economy on the recent slowdown in health spending. Kaiser Family Foundation.Find this resource: Lindert, P., & Williamson, J. (2016). Unequal gains: American growth and inequality since 1700. Princeton, NJ: Princeton University Press. The city tables are available online at the Global Price and Income History Group website. under American incomes ca. 1650–1870 data, Occupation codes and counts, c. 1774.Find this resource: Lough, W. H. (1935). High-level consumption, its behavior; its consequences. New York, NY: McGraw-Hill.Find this resource: Mays, R., (2007). Origins development, and passage of Medicare’s revolutionary prospective payment system. Journal of the History of Medicine and Allied Sciences, 62(1), 21–55.Find this resource: Meara, E., White, C., & Cutler, D. M. (2004). Trends in medical spending by age, 1963–2000. Health Affairs, 23,176–183.Find this resource: Millard, C. (2011). The destiny of the republic: A tale of madness, medicine, and the murder of a president. New York, NY: Doubleday.Find this resource: Mokyr, J. (1998). Induced technical innovation and medical history. Journal of Evolutionary Economics, 8, 119–137.Find this resource: Morgan, D., & Astolfi, R. (2013). Health spending growth at zero: Which countries, which sectors are most affected? (OECD Health Working Paper No. 60). Paris, France: Organisation for Economic Co-operation and Development.Find this resource: Morrisey, M. A. (2008). Health insurance. Chicago, IL: Health Administration Press.Find this resource: Moses, H., Matheson, D. H. M., Carins-Smith, S., George, B. P., Palisch, C., & Dorsey, E. R. (2015). The anatomy of medical research: US and international comparisons. JAMA, 313, 174–189.Find this resource: Moses, H., Matheson, D. H. M., Dorsey, E. R., George, B. P., Sadoff, D., & Yoshimura, S. (2013). Anatomy of health care in the U.S. JAMA, 310, 1947–1963.Find this resource: Mount Vernon. (2017). Death of George Washington. George Washington’s Mount Vernon (website).Find this resource: Muller, U., & Watson, M. W. (2017). Long-run covariability (NBER working paper No. 23186). National Bureau of Economic Research. Cambridge, MA.Find this resource: Murphy, K. M., & Topel, R. H. (2006). The value of health and longevity. Journal of Political Economy, 114, 871–904.Find this resource: Murray, J. E. (2007). Origins of American health insurance. New Haven, CT: Yale University Press.Find this resource: National Center for Health Statistics. (2016). Health, United States. Hyattsville, MD.Find this resource: Newhouse, J. P. (1992). Medical care costs: How much welfare loss? Journal of Economic Perspectives, 6(3), 3–21.Find this resource: Numbers, R. L. (1979). The third party: Health insurance in America. In Vogel & Rosenberg (1979), 188–199.Find this resource: Organisation for Economic Co-operation and Development. (1977). Public Expenditure on Health. Paris: OECD.Find this resource: Organisation for Economic Co-operation and Development. (1985). Measuring Health Care 1960–1983: Expenditures, Costs and Performance. Paris: OECD.Find this resource: Organisation for Economic Co-operation and Development. (2016). OECD Health Data 2016. Paris, OECDFind this resource: Osler, W. (1892). The principles and practice of medicine. Norwalk, CT: Appleton-Century-Crofts. This textbook was revised and updated in 1909, and a modernized edition published in 1984.Find this resource: Osler, W. (1922). The evolution of modern medicine; a series of lectures delivered at Yale University April 1913. New Haven, CT: Silliman Foundation.Find this resource: Paulson, G. (2006). Death of a president and his assassin: Error in their diagnosis and autopsies. Journal of the History of the Neurosciences, 2, 77–91.Find this resource: Peden, E., & Freeland, M. (1998). Insurance effects on U.S. medical spending (1960–1993). Health Economics, 7, 671–687.Find this resource: Perkins, B. P. (1998). Economic organization of medicine and the committee on the costs of medical care. American Journal of Public Health, 88, 1721–1726.Find this resource: Ramey, V. (2016). Macroeconomic shocks and their propagation. NBER working paper No. 21978. National Bureau of Economic Research. Cambridge, MA.Find this resource: Reed, L. S. (1940). Costs and benefits under prepayment medical service plans. Social Security Bulletin, 3(3), 13–26.Find this resource: Reed, L. S. (1966). Private health insurance: Coverage and financial experience, 1940–1966. Social Security Bulletin, 29(11), 3–22.Find this resource: Reed, L. S., & Hanft, R. (1966). National health expenditures, 1950–64. Social Security Bulletin, 29(1), 3–19.Find this resource: Reed, L. S., & Rice, D. P. (1964). National health expenditures: Object of expenditures and source of funds, 1962. Social Security Bulletin, 27(8), 11–21.Find this resource: Reinhardt, U. (2003). Does the aging of the population really drive the demand for health care? Health Affairs, 22, 627–639.Find this resource: Rice, D. P., & Cooper, B. S. (1970). National health expenditures, 1929–68. Social Security Bulletin, 33(1), 3–20.Find this resource: Roehrig, C., Turner, A., Hughes-Cromwick, P., & Miller, G. (2012). When the cost curve bent: Pre-recession moderation in health care spending. New England Journal of Medicine, 367, 590–593.Find this resource: Rosen, G. (1944). The specialization of medicine, with special reference to ophthalmology. New York, NY: Arno.Find this resource: Rosen, G., & Rosenberg, C. E. (1983). The structure of American medical practice, 1875–1941. Philadelphia: University of Pennsylvania Press.Find this resource: Rosenberg, C. E. (1987). The care of strangers: The rise of America’s hospital system. New York, NY: Basic.Find this resource: Rubinow, I. M. (1913). Social insurance. New York, NY: Henry HoltFind this resource: Rutkow, I. (2006). James A. Garfield. New York, NY: Macmillan.Find this resource: Ryu, A. J., Gibson, T. B., McKellar, M. R., & Chernew, M. E. (2013). The slowdown in health care spending 2009–2011. Health Affairs, 32, 1–5.Find this resource: Scofea, L. A. (1994, March). Development and growth of employer-provided health insurance. Monthly Labor Review, 3–10.Find this resource: Seale, J. R. (1959). A general theory of national expenditure on medical care. The Lancet, 274, 555–559.Find this resource: Selden, T. M., Levit, K., Cohen, J. W., Zuvekas, S. H., Moeller, J. F., McKurick, D., & Arnett, R. (2001). Reconciling medical expenditure estimates from the MEPS and NHA, 1996. Health Care Financing Review, 23(1), 161–178.Find this resource: Sensenig, A. L., & Wilcox, E. (2001). National health accounts, national income and product accounts reconciliation: Hospital care and physician services. In D. M. Cutler & E. R. Berndt (Eds.), Medical care output and productivity (pp. 271–302). Chicago, IL: University of Chicago Press.Find this resource: Shryock, R. H. (1947a). The American physician in 1846 and 1946. JAMA, 134, 417–424.Find this resource: Shryock, R. H. (1947b) American medical research: Past and present. New York, NY: Commonwealth Fund.Find this resource: Silver, N. (2012). The signal and the noise: Why so many predictions fail—but some don’t. New York, NY: Penguin Press.Find this resource: Sims, C. (1996). Macroeconomics and methodology. Journal of Economic Perspectives, 10(1), 105–120.Find this resource: Smith, D. B. (2015). Civil rights and Medicare: Historical convergence and continuing legacy. In A. B. Cohen, D. C. Colby, K. A. Wailoo, & J. Zelar (Eds.), Medicare and Medicaid at 50: America’s entitlement programs in the age of affordable care (pp. 21–38). New York, NY: Oxford University Press.Find this resource: Smith, S., Freeland, M., Heffler, S., McKusick, D., & the Health Expenditures Projection Team. (1998). The next ten years of health care spending: What does the future hold? Health Affairs, 17, 128–140.Find this resource: Smith, S., Newhouse, J. P., & Freeland, M. (2009). Income, insurance, and technology: Why does health spending outpace economic growth? Health Affairs, 28, 1276–1284.Find this resource: Society of Actuaries. (2018). Modeling long-term healthcare cost trends (multiple years, multiple reports with technical manual).Find this resource: Sobek, M. (2001). New statistics on the U.S. labor forces, 1850–1990. Historical Methods, 34, 71–87.Find this resource: Starr, P. (2008). Social transformation of American medicine: The rise of a sovereign profession and the making of a vast industry. New York, NY: Basic.Find this resource: Stevens, R. (1971). American medicine and the public interest. New Haven, CT: Yale University Press.Find this resource: Stevens, R. (1998). In sickness and in wealth: American hospitals in the twentieth century. Baltimore, MD: Johns Hopkins University Press.Find this resource: Stigler, G. J. (1954). The early history of empirical studies of consumer behavior. The Journal of Political Economy, 62(2), 95–113.Find this resource: Stigler, G. J. (1956). Trends in employment in the service industries. National Bureau of Economic Research General Series 59. Princeton, NJ: Princeton University Press.Find this resource: Stone, R. (1997). The accounts of society. American Economic Review, 87, 17–29.Find this resource: Swain, D. C. (1962). The rise of a research empire: NIH 1930 to 1950. Science, 138, 1233–1237.Find this resource: Thomasson, M. A. (2002). From sickness to health: The twentieth-century development of U.S. health insurance. Explorations in Economic History, 39, 233–253.Find this resource: U.S. Government Accountability Office. (1976). History of the rising costs of Medicare and Medicaid and attempts to control these costs 1966–1975. Washington, DC: U.S. Government Printing Office.Find this resource: Vogel, M. (1980). The invention of the modern hospital, Boston 1870–1930. Chicago, IL: University of Chicago Press.Find this resource: Vogel, M., & Rosenberg, M. (1979). The therapeutic revolution: Essays in the social history of American medicine. Philadelphia: University of Pennsylvania Press.Find this resource: Waldo, D. R., Sonnefeld, S. T., McKusick, D. R., & Arnett, R. H., III. (1989) Health expenditures by age group, 1977 and 1987. Health Care Financing Review, 10(4), 111–120.Find this resource: Wallenborn, W. M. (1997). George Washington's terminal illness: A modern medical analysis of the last illness and death of George Washington. Washington Papers (website). University of Virginia, Charlottesville.Find this resource: Warner, J. H. (2014). The therapeutic perspective: Medical practice, knowledge, and identity in America, 1820–1885. Princeton, NJ: Princeton University Press.Find this resource: Washington, A. E., Coye, M. J., & Feinberg, D. T. (2013). Academic medical centers and the evolution of the health care system. JAMA, 310, 1929–1930.Find this resource: Weinfeld, W. (1951, July). Income of physicians, 1929–1949. Survey of Current Business, 9–26.Find this resource: Weisbrod, B. A. (1991). The health care quadrilemma: An essay on technological change, insurance, quality of care, and cost containment. Journal of Economic Literature, 29, 523–552.Find this resource: White, C., & Ginsburg, P. (2012). Slower growth in Medicare spending: Is this the new normal? New England Journal of Medicine, March 7, 2012, 366, 1073–1075.Find this resource: White, H., & Granger, C. W. J. (2011) Consideration of trends in time series. Journal of Time Series Econometrics, 3(1), art. 2.Find this resource: Williams, P. (1932). The purchase of medical care through fixed periodic payment. New York, NY: National Bureau of Economic Research.Find this resource: Williamson, J. G. (1967). Consumer behavior in the nineteenth century: Carroll D. Wright’s Massachusetts workers in 1875. Explorations in Entrepreneurial History, 2nd ser., 4(2), 98–135.Find this resource: Wright, C. D. (1894). Analysis and index of all reports issued by Bureaus of Labor Statistics prior to November 1, 1892. Washington, DC: U.S. Government Printing Office.Find this resource: ## Notes: (1.) Table 1 and Figure 1: National Health Expenditure shares 1960–2016 are from the current CMS OACT NHE Historical Series (CMS, 2017a). Other years are as follows: 1950 is from Reed and Hanft (1966) linked to the current series at 1960 by adjustment x1.02 using relative values; 1940 is a linear extrapolation between 1950, and 1930 is an author estimate based of a variety of sources, including Committee on the Costs of Medical Care (1932), Seale (1959), and Cooper et al. (1973), with adjustment for distortion created by sharp recession in 1929 (see Getzen, 2017, for details). Lough (1935) is used to establish cagr trends to extrapolate back from 1930 to 1910. Based on census employment trends, Getzen extrapolates backward using cumulative annual-growth-rate estimates of 0.8% for 1900–1910, 0.4% for 1880–1900, and 0.2% for 1850–1880, and extends backward at the same 0.2% rate to 1776, relying on the Lindert city occupational totals as an endpoint. Real GDP per capita: Bureau of Economic Analysis (2017) Current-Dollar and “Real” Gross Domestic Product for 1929–2016 and online Table Ca11 in Historical Statistics of the United States, Millennium Edition (Carter et al., 2006); for 1790–2000 extrapolated back to 1776 using the 0.73% annual growth rate from 1790–1850. Resident population in millions from Table Ca14 (identical to Aa7). Urbanization Table-Aa699-715 for 1790–1999 and “percent urban” for 2000 and 2010 from Bureau of the Census (2018.) Physicians from U.S. Decennial Census occupations 1850–1990 in Carter et al. (2006) Tables Ba1218 & Ba1222; for 1970–2000, from Health, United States, 1993, tables 108 and 109; and Health, United States, 2015 table 83. Health Employment: 1990–2016, from BLS “Current Employment Statistics Survey” data series CEU6562000101; and for 1960–1990, the comparable Standard Industrial Classification (SIC) codes 808 “Health” series from BLS Handbook of U.S. Labor Statistics, 2003, table 2–1 (adjusted x1.07 to link with the current North American Industrial Classification (NAIC) code series). For earlier years, U.S. Decennial Census 1850–1970 (Carter et al., 2006) Table Ba1033–1439 (occupations categorized as “health employment” by author), see also Alba Edwards (1943), Comparative Occupational Statistics for the United States, 1870 to 1940. Health-employment estimates in the older BLS SIC series are somewhat smaller than in the current NA Industrial Classification code series, which began in 1990, and also differ from census occupational totals for 1960–1990. Staff per MD is total health employment (minus physicians) divided by number of physicians. Note that various vintages or versions the “same” data series often show different values for the same year. Life Expectancy: For 1900–2016: National Center for Health Statistics (NCHS) Vital Statistics Reports 66 no. 4, “United States Life Tables, 2014” August 2017, and NCHS data brief no. 293, “Mortality in the United States, 2016.” For 1800–1880, see J. D. Hacker (2010), Decennial life tables for the white population of the United States 1790–1900, Table 8.. see also M. R. Haines (2006), “Vital Statistics” and Carter et al. (2006), Table Ab644, “Life Expectancy at Birth 1850–1998.” (2.) There are many hundreds of studies of health expenditure growth in the United States, and most are excluded from this review. Time series of fewer than 20 years are too short to provide robust evidence of changes in long-run trends that could meaningfully be related to changes in policy, events, technology, organization, or macroeconomic conditions. Many studies use geographic subdivisions in panels that do not correspond to national trends. Also excluded are studies limited to determination of factor contributions rather than shifts in trends. The fourteen studies listed in Table 2 include almost every study of long-run trend changes, with particular focus on studies from before 1990 that may not be as well-known or readily available to most researchers. Also excluded are the many cross-national time series studies that do provide insight into U.S. national trend differences but were deemed outside the scope of this article to maintain a focus on the U.S. trends. It must be recognized, however, that the compilation of comparative international data sets by Jean-Pierre Poulier at the Organisation of Economic Co-operation and Development (1977, 1985), the World Health Organization, and the World Bank have transformed the understanding of U.S. trends. (3.) Measuring expenditure shares relative to GDP is uncommon before 1970. GNP was the more typical aggregate at mid-century, and prior to the 1930s, there was no consistent set of accounts for national income that provided a single measuring rod. Expenditure shares in earlier periods must be measured relative to aggregate or average spending or incomes for a specific group of persons. See Bureau of Economic Analysis (2017) NIPA Handbook: Concepts and methods of the U.S. National Income and Product Accounts: Coile (2014); Fogel, Fogel, Guglielmo, and Grotte (2013); and Stone (1997), (4.) Weisbrod (1991, p. 529) also discusses how size affects interaction: Switzerland is small, so research and medical spending are independent, whereas the United States is large enough to experience general equilibrium feedback between research and NHE. (5.) This is the only detailed analysis of trend changes by the OACT team, although the annual NHE estimates published in the Social Security Bulletin (1960 to 1977), Health Care Financing Review (1978 to 2002) and Health Affairs (1988 to present) often include brief comments about trends or changes in trend. (6.) See Bernard et al. (2012); Hartman, Kornfeld, and Catlin (2010); Selden et al. (2001); Sensenig and Wilcox (2001). (7.) Mokyr (1998) has stated, “The history of medicine, as viewed from the point of view of the technological historian, shows remarkably little progress of any significance before 1800. Indeed, it could be argued that the ability of mankind to understand, avoid, let alone cure diseases by 1850 was little better than it had been at the time of Galen” (p.121). See Bynum (1994); Kett (1968); King (1991); Osler (1922); Rosen and Rosenberg (1983); Shyrock (1947a); Warner (2014). (8.) See Anderson (1990); Consoli and Mina (2009); Dranove (2009); Rosen, (1944); Rosenberg (1987); Starr (2008); Stevens, (1971, 1998); Vogel (1980); Vogel and Rosenberg (1979). (9.) The Bureau of Labor Statistics also published a compendium of 19 workmen’s budgets from 1914 to 1919 indicating medical costs averaging 2.9% of working family expenditures, but with a wide range (1.5%–5%; Bureau of Labor Statistics, 1920). These budget studies did not rely on regular survey data, frequently using estimates by the authors or placing a standard amount ($70) to fill in the gaps. In a later effort to construct category weights for a consumer price index, in 1917–1918, the BLS surveyed 12,096 urban families and determined that average medical expenditures were $64, or 4.7% of total consumption (Bureau of Labor Statistics, 1924). The growth rate of employment for 1910–1930 shown in Table 1 is based on census measures of persons with health occupations rather than employment per se. (10.) Williams (1932, p. 291) stated, “The number of American trade union members who secure medical, surgical or hospital care as a benefit . . . is negligible.” See also Klem (1942); and Reed (1940, 1966). (11.) The rate of increase in real per-capita health spending for 1930–1955 (3.7%) is slightly larger than for 1910–1930 (3.5%) even though the rate of excess medical cost growth is much lower (0.4% vs. 1.4%) because per-capita incomes grew so much more rapidly during the later period. (12.) Table 5. The American Hospital Association has tabulated data on the number, admissions, days of care, revenues, expenditures, and full-time-equivalent employees of hospitals annually since 1946; the most recent data is available in the annual AHA Trendwatch Chartbook online at https://www.aha.org/guidesreports/2018-05-22-trendwatch-chartbook-2018. The American Medical Association tabulated slightly different data for selected years from 1909 to 1953, so the estimates for years prior to 1946 are not exactly comparable. Most of the data is available in Historical Statistics of the United States, Millennium Edition (Carter et al., 2006), in tables Bd83-Bd-276, and the recent data is also available in the National Center for Health Statistics’ Health, United States for various years. According to The Lancet (1874) 2,717–720, the cost of private hospital beds was$2 to $5 per day in 1874. (13.) Table 6. Provider percentages for 1960–2016 are from Centers for Medicare and Medicaid Services (2017a); for 1955, from Reed and Hanft (1966); for 1929, from Committee on the Costs of Medical Care (1932); and for 1910, Lough (1935). Categorizations for 1910 and 1929 prior to the development of NHEAs are not fully comparable with recent data. (14.) Table 7. Payer percentages for 1960–2015 are from Centers for Medicare and Medicaid Services (2017a) NHEA; for 1950, from Reed (1966) and Reed and Hanft (1966); and for 1929, from Committee on the Costs of Medical Care (1932). Coverage percentages for 2015 are from Health, United States, 2016 (National Center for Health Statistics, 2016); for 1959–2007, from Cohen et al. (2009); and for prior years, from Numbers (1979). See also Cunningham and Cunningham (1997); Morrisey (2008); Reed (1966); Anderson, Sinai, and Dollar (1946); Scofea (1994); and Thomasson (2002). (15.) “Price” is a problematic term for healthcare services. Most transactions involve complex third-party payments that often include multiple indirect subsidies. The first distortion mentioned here relates to the fact that some insurance premiums and expenditures amount to untaxed compensation. The second is the great disparity between listed charges and the amount that is usually paid after contractual adjustments. (16.) CMS (2017a) Table 5. Note that that the taxes supporting Medicare and worker’s compensation are categorized as being sponsored by business and households rather than government. In addition, the tax subsidy for private health insurance is about$300 billion, 9% of total national health expenditures (Congressional Budget Office, 2017).

(17.) See Blumenthal, Stremikis, and Cutler (2013); Chernew (2014); Cutler and Sahni (2013); Levitt et al. (2013); Roehrig, Turner, Hughes-Cromwick, and Miller (2012); Ryu, Gibson, McKellar, and Chernew (2013); and White and Ginsburg (2012).

(18.) From 1900 to 1940, 34 of the 42 Nobel Prizes awarded in medicine came from Europe and just four from the United States, the first of which was in awarded in 1933. Since 1940, the majority of the prizes have been awarded to scientists from the United States.

(19.) OECD Health Data for 2016 (Organisation for Economic Co-operation and Development, 2016), and previous years. See also Gerdtham and Jönsson (2000); Gerdtham, Sogaard, Andersson, and Jonsson (1992); Horenstein and Santos (2016). Note that OECD health expenditures data for the United States differ slightly from the CMS OACT NHE estimates. See also Garber and Skinner (2008).

(20.) Seale (1959) appears to have been the first to report lags in national health expenditures; and Getzen (1985), the first estimates of lag coefficients. Complexity, information asymmetry, contractual incompleteness, institutional ossification, and moral entanglements may all play a role in making the health sector slow to respond (Arrow, 1963; Frandsen, Brigham, & Rebitzer, 2017). The fact that most consumption is less volatile than income and tends to respond with a lag that displaces and dampens business cycles has long been noted (Cochrane, 2016; Ramey, 2016). In his Nobel address, Angus Deaton (2016) noted his early research showing how unexpected changes in inflation create delays in the consumer purchasing response (Deaton, 1977). Some of the challenges in determining temporal dynamics are addressed by Sims (1996); Cochrane (1994); Muller and Watson (2017); Ramey (2016); and White and Granger (2011); See also Getzen (2000, 2014); and Kendix and Getzen (1994).

(21.) Table 9. Tracking the distribution of health expenditures across age groups over time is complicated by inconsistencies in measurement, particularly regarding the inclusion of institutional long-term care and home health. Therefore Table 9 relies most heavily on two sources that use consistent methods to cover extended spans of time, Meara, White, and Cutler (2004) for 1963–2000, and Centers for Medicare and Medicaid Services (2017c), NHE Age and Gender Tables for 2002–2012. All of the other sources have been reviewed and a spreadsheet including data from all sources for all available years can be obtained from the author. The 1953 estimate from Cutler and Meara (1997) is based on the Non-partisan and Objective Research Center at the University of Chicago (NORC) survey conducted by Odin Anderson (1963) and therefore differs somewhat from later estimates, but was included here to provide a better sense of spending patterns for the decades prior to the development of Medicare and the expansion of insurance coverage. Other sources consulted include Cooper et al. (1973); Fisher (1980); Hartman, Catlin, Lassman, Cylus, and Heffler (2008); Keehan, Lazenby, Zezza, and Catlin (2004); Lassman, Hartman, Washington, Andrews, and Catlin (2014); and Waldo, Sonnefeld, McKusick, and Arnett (1989).

(22.) Table 10. Note that the last column of Table 9, percentage of spending age 65+, applies to personal health spending. In order to decompose NHE, the 15% not directly attributable to individual patients (research, administration, overhead, construction, public health) must be allocated proportionately (as is done here) or by some other method.

(23.) The 65+ group accounted for 20% of expenditures in 1963 and 43% in 1987 as the health share of GDP increased from 5.4% to 10.6%. Hence it is not unreasonable to extrapolate that of the +5.2% incremental increase in the health share of GDP, +3.4% (about two thirds) could be attributed to the elderly, rising from 1.1% of GDP (20% of 5.4%) to 4.5% (43% of 10.6%).

(24.) The CBO projection methodology is explained in the annual Long Term Budget Outlook (Congressional Budget Office, 2017) posted on the CBO website. The CMS has posted a projection methodology paper (Centers for Medicare and Medicaid Services, 2017a) and provides its’ most recent publication on the CMS NHE website and in a Health Affairs article (Keehan et al., 2017). The SOA projection methodology is described in a technical manual posted along with the most recent model update on the SOA website (Society of Actuaries, 2018).

(25.) Keehan et al. (2017) projects the health share of GDP to be 19.9% of GDP in 2025. An assessment of the accuracy of CMS 10-year projections over the last two decades by Getzen (2016b) indicates cumulative error and drift of approximately ±1% per year (±10% per decade). Estimates of uncertainty in forecasts for more than a decade are problematic. Silver (2012) describes weather forecasters using massive models that have become quite good for making forecasts up to 10 days into the future; but any forecast longer than that must default to very simple historical averages (and, to make climate-change projections over centuries or millennia, must use entirely different kinds of models). Analogously, inertia and business-cycle lags within the healthcare system allow for fairly good forecasts up to 10 years, but for any longer forecasts, must default to simple long-run historical averages—raising the question of whether the average should be for the last 20, 50, or 200 years and if it should be weighted. The judgment made for the SOA model is that the 25-year average provides the best trade-off between currentness and length.