Good health is a key part of our quality of life. In this entry we focus on healthcare – one of the most important inputs to protect and improve health. There are many other factors affecting health, and you can read more about some of them in our entries about health.
Publicly funded healthcare is a legacy of the Age of Enlightenment.1 The first examples of legislation on health insurance date back to the late 19th century.2 Data from these early systems shows that healthcare expenditure only began rising several years after the expansion of insurance coverage, with the discovery of powerful new treatments.3
The impact that scientific developments had on healthcare expenditure is epitomized in the U.S. experience: in recent decades, as treatment possibilities expanded rapidly, expenditure on healthcare increased – in any way you want to measure it: private and public, both per capita and as a share of gross domestic product. This occurred without major changes in insurance coverage and had two important consequences: (i) the U.S. currently spends more government money per person on healthcare than many countries that fund universal programs, and (ii) spending is so concentrated that the top 1% of spenders account for more than 20% of total healthcare expenditure.4
Global expenditure on healthcare as a share of world income has been increasing, steadily but slowly over the course of the last couple of decades. In the background, however, there has been substantial cross-country heterogeneity, both in levels and trends. Regionally, high-income countries spend – and have been spending – a much larger share of their income on healthcare than low-income countries (about twice as much). Moreover, in contrast to high-income countries, in low and middle-income countries the public share of healthcare funding is much lower – although it has been catching up – and the role of out-of-pocket expenditures is much higher (above 50% of total expenditure in many countries).5
Healthcare financing in developing countries in the 21st century has been largely shaped by the flow of resources channeled through development assistance. These flows – which saw a steep increase after the introduction of the Millennium Development Goals – account for around 0.7% of the resources spent by high-income countries on healthcare. Although this may seem small in proportion to the national commitments of rich countries, for low-income countries at the receiving end of the transfers, these resources are substantial; in Sub-Saharan Africa they finance more than 25% of total expenditure on healthcare.6 This implies that development assistance for health, if suitably targeted and managed, has the potential of drastically reducing inequality in health outcomes: the robust empirically observed relationship between health outcomes and healthcare spending is indicative of large returns to healthcare investments, particularly at low levels of baseline expenditure.7
All our interactive charts on Financing Healthcare
Nowadays healthcare is commonly considered a ‘merit good’ – a commodity which is judged that an individual or society should have on the basis of need rather than ability and willingness to pay. This view, partly grounded on the recognised positive externalities of healthcare consumption, is perhaps most visibly materialized in the fact that access to healthcare is currently a constitutional right in many countries.8
However, just a few generations ago the situation was very different. In fact, during the Middle Ages health was considered a matter of destiny across most of Western Europe; it was only afterwards, under the influence of Mercantilism and the Enlightenment, that this view started changing and public authorities increased their ambitions concerning the promotion of public health.9 Sundin and Willner (2007) say that “[g]enerally, before the era of the Enlightenment, it was thought that health was God’s gift and disease and death was His punishment for the sins of an individual, the congregation, the whole nation or its rulers. Hence, to live a decent life in accordance with His will and repenting one’s sins were considered the most effective preventive measures against illnesses”.10
To show this development we have produced a long-run dataset with estimates of government expenditure on healthcare as a percent of gross domestic product (GDP) for a selection of high-income countries, going back to 1880.
As the data show, in 1880 government health spending was below 1% of GDP in all countries; but this started changing quickly in the first half of the 20th century and by 1970 government spending on healthcare was above 2% of GDP in all these countries.
The steeper increase in public expenditure on healthcare observed in European countries after the Second World War is largely due to the fact that medicine had major breakthroughs during the second half of the 20th century – beginning, notably, with the discovery and use of penicillin and other antibiotics. Before these scientific developments took place, the main component of healthcare was not treatment but income insurance, an insurance paying benefits to those who were unable to work due to poor health. In fact public health insurance for workers was already substantial in a number of European countries before the Second World War. The visualization presents public health insurance coverage as percent of labour force for a number of European countries (missing observations in this graph reflect lack of public insurance programs for the corresponding year/country). Tanzi and Schuknecht (2000) – the source for the data in this figure – note that by 1929 all the eleven European countries shown in this visualization had at least voluntary insurance schemes; and that by 1935 half their labor force was covered by health insurance.
Having introduced universal access to healthcare in 1948 through the National Health Service, the United Kingdom is a particularly interesting instance to study in detail.11 The visualization shows that the costs of this universal-access system grew more in the first decade of the 21st century, than they did in the first two decades immediately after its inception.
The lion’s share of the above-mentioned historical increase in healthcare expenditure in the U.S. took place without an underlying increase in the share of people who were covered by health insurance. The visualization supports this; it presents a plot of coverage rates by type of plan (particular care should be taken when reading this graph, since insurance plans are not mutually exclusive; this means that those covered by ‘private’ and ‘government’ plans add up to more than those covered by ‘any plan’). As it can be seen, total health insurance coverage remained virtually constant at around 85% for decades, while private and public healthcare expenditure increased continuously over the same period. Interestingly, this data also shows an apparent change in coverage trends after 2012, when the ‘Affordable Care Act’ started kicking in. The most recent Gallup estimates (fourth quarter of 2015) place total coverage at 88.1%.
The fact that insurance coverage remained stable while healthcare spending was increasing rapidly due to major improvements in treatment possibilities during the 20th century, implied that healthcare expenditure in the U.S. grew highly concentrated. The graph, produced by the National Institute for Health Care Management (NIHCM), shows the cumulative distribution of healthcare spending in the U.S., using data on personal expenditures during the year 2009, across the entire ‘non-institutionalized civilian population’.
The source of the data for this visualization is the Medical Expenditure Panel Survey – a nationally representative longitudinal survey that collects information on healthcare utilization and expenditure, health insurance, and health status, as well sociodemographic and economic characteristics for civilian non-institutionalized population. According to the source, the data refers to ‘non-institutionalized civilian population’, in the sense that it excludes care provided to residents of institutions, such as long-term care facilities and penitentiaries, as well as care for military and other non-civilian members of the population. The data corresponds to ‘personal healthcare services’, in the sense that they exclude administrative costs, research, capital investments and many other public and private programs such as school health and worksite wellness.
This graph should be read similarly to a Lorenz curve: the fact that the cumulative distribution of spending bends sharply away from the 45% degree line is a measure of high inequality (this is the intuition of the Gini coefficient that we discuss in our income inequality data entry). As it can be seen, the top 5% of spenders account for almost half of spending, and the top 1% account for more than 20%. While some concentration in expenditure is to be expected when looking at the distribution across the entire population – because it is in the nature of healthcare that some individuals, particularly those older and with complicated health conditions, will require large expenditure –, these figures seem remarkably large. They suggest important inequality in access, over and above inequality in need. Indeed, the publisher of the graph notes that a report from the Medicare Payment Assessment Commission shows that personal spending for individuals covered by Medicaid is somewhat less concentrated than for the population as a whole.12
Cumulative distribution of personal healthcare spending in the U.S., 2009 – NIHCM (2012)13
We have already pointed out that European countries pioneered the expansion of healthcare insurance coverage in the first half of the twentieth century. The visualization, from the Human Development Report (2014), places the achievements of these countries in perspective. Specifically, the graph plots healthcare protection coverage for a selection of countries during the period 1920-2010. As we can see, France, Austria and Germany increased healthcare coverage in the years 1920-1960, while Spain, Portugal and Greece did it later, in the years 1960-1980. Interestingly, however, this graph also shows some notable examples of countries that expanded healthcare coverage much later, but much more quickly. In particular, China, Rwanda and Vietnam built health protection systems in the 21st century, almost from scratch, achieving near universal coverage in only a decade. These examples show that healthcare protection can be expanded very quickly, and not only at low baseline levels of coverage.
The map shows estimate of healthcare insurance coverage across the world. Insurance coverage here includes affiliated members of health insurance, as well as the population having free access to healthcare services provided by the government.
As can be seen, all high-income countries have virtually full health insurance coverage, with the notable exception of the US. And similarly, many middle-income countries, such as Brazil and China, also have very high levels of coverage.
In most low-income countries coverage remains a challenge; but here there are also important exceptions. Gambia, Rwanda, and Vanuatu, for example, all have higher health protection coverage than the US.
In this scatter plot you can compare health insurance coverage figures by national GDP levels.
The visualization – produced by the Institute for Health Metrics and Evaluation (IHME) – presents public health expenditure from autonomous sources in absolute terms (billions of 2011 U.S. dollars) for developing countries.15
These numbers correspond to public healthcare expenditure after removing funds provided directly to the developing countries by development assistance partners. As we can see, healthcare spending from autonomous sources has increased substantially in these countries. Below we discuss development assistance for health in more detail.
Public health expenditure in developing countries after removing funds from development assistance – Figure 44 in IHME (2013)16
Despite significant cross-country heterogeneity in health expenditure, all countries spend less than twenty percent of gross domestic product (GDP) on healthcare. Most countries spend between 5-12 percent of GDP.
This is largely at odds with public perception of healthcare spending—all over the world, people grossly overestimate actual healthcare spending.
The chart shows this using data from the Perils of Perception Survey (Ipsos MORI, 2016). On the vertical axis we see the average that survey respondents guess is spent on health every year, as share of GDP. And on the horizontal axis we see estimated actual expenditure (also as share of GDP).
As we see, all countries included in the survey lie above a hypothetical parity diagonal (i.e. a line with slope one), meaning that people everywhere think they spend more on healthcare than they actually do. Those countries which lie closest to the parity diagonal have a more accurate perception of health spending—in particular, Poland, Russia and Hungary overestimated spending by four percentage points.
In some cases disparities are huge. In Indonesia, the country furthest away from the parity diagonal, respondents thought spending was 39 percent of GDP, while the actual figure is only three percent. As an average across all countries included in this survey, respondents thought health spending amounted to 21 percent of GDP—an overestimation of 13 percentage points.
As we discuss below, estimates of actual healthcare spending are subject to an important margin of error. However, perceptions are so far from official figures that it is hard to reconcile the discrepancies through measurement error. The most likely explanation is that people have skewed perceptions.
The Millennium Development Goals have been associated with major increases in global health financing flows, particularly for the health focus areas explicitly targeted (fight against child mortality, maternal mortality, HIV/AIDS, malaria, and tuberculosis). An important part of these financing flows occur under the label of development assistance. The Institute for Health Metrics and Evaluation (IHME) defines development assistance for health as all financial and in-kind contributions provided by global health channels to improve health in developing countries (including grants, as well as concessionary loans, provided with no interest or at a rate significantly lower than the going market rate).
The IHME reports that since the formation of the Millennium Development Goals, $227.9 billion in development assistance has targeted these health focus areas. These flows account for around 0.7% of the resources spent by high-income countries on healthcare. Although this may seem small in proportion to the national commitments of rich countries, for low-income countries at the receiving end of the transfers, these resources are substantial; in sub-Saharan Africa they finance more than 25% of total expenditure on healthcare17
The report Financing Global Health 2014, produced by the IHME, provides a detailed account of this source of funding for healthcare, and how it has changed over the last two decades. The data can be accessed from IHME’s website, where there are also some excellent interactive visualizations. The chart shows the evolution of development assistance for health by source, over the last fifteen years. As it can be seen, these funds increased sharply in the period 2000-2010, but have plateaued since. Considering that this source of funding has a larger weight in those countries with the lowest income, the recent change in trend is particularly problematic for the poorest.
Sources of development assistance for health, 1990-2014 – IHME18
The visualization uses aggregate 2000-2012 figures to show the relationship between sources of development assistance funds, and the corresponding channels and recipient regions. The main features here are the weight of the U.S. as a source-channel, and the sub-Saharan Africa region as a recipient.
Flows of development assistance for health, from source to channel to recipient region, cumulative 2000-2012 – Figure 11 in IHME (2014)19
In many countries an important part of the private funding for healthcare takes the form of ‘out-of-pocket’ spending. This refers to direct outlays made by households, including gratuities and in-kind payments, to healthcare providers. The visualization presents out-of-pocket expenditure on healthcare by country (as percent of total healthcare expenditure). As it can be seen, in high-income countries these outlays tend to account for only a small fraction of expenditure on healthcare (e.g. France, where the share was always below 8% in the entire series 1995-2013); while in low-income countries, they account for the majority of funding (e.g. Afghanistan, where the share of out-of-pocket expenditure reached 87.7% in 2002). Many countries have followed a clear path in the direction of reducing this type of expenditures (particularly in the developing world), yet some countries have moved in the opposite direction (Russia is a notable case in point, with a threefold increase in the share of out-of-pocket expenditure in the last decade).
This relationship between income and reliance on out-of-pocket health expenditures is further shown in the chart. Here, we see the share of out-of-pocket expenditure as a percentage of total healthcare expenditure (on the y-axis) versus gross domestic product (GDP) per capita (which has been PPP-adjusted) on the x-axis. Overall, we see that these outlays tend to account for a smaller fraction of overall health spending in higher-income countries versus low-income nations.
Levels of income can therefore affect two aspects of healthcare financing: the magnitude of total health expenditure, in addition to the source of such funding. This is shown in the chart, which presents the best-fit global trends of total per capita health expenditure (the blue line); share of out-of-pocket expenditure in total spending (orange line); and the share that comes from external (donor) funding (green line). Here, external funding refers to economic resources from non-resident units channeled towards healthcare (whether explicitly labelled so, or not), through the government or private sector.20 This data is measured relative to the average GDP per capita (shown on a log-scale x-axis).21
The first point concerns the relationship between per capita health expenditure and income. Overall, as countries get richer, per capita expenditure on healthcare tends to increase (this has been discussed earlier in the entry, and is shown here by the rising blue trend line).
The second key point concerns the source of healthcare funding. In the chart we observe that as per capita income increases, the share of both out-of-pocket outlays and external donor funding decreases. As the contribution of these sources decline, typically the share from public (i.e. governmental) funding increases.
However, we also notice that out-of-pocket and external funding contributions decline at different rates. On average, external donor funding decreases at a lower income level than out-of-pocket outlays, and shows a significantly steeper decline. External donor funding is often the dominant source of healthcare spending for the poorest, but is quickly replaced by other sources as those on very low incomes move towards low- and lower-middle incomes.
For poor countries with a per capita GDP of less than 500 US$ per year, donor funding accounts for approximately 45 percent of health expenditure, on average. This drops to 34 percent for countries up to 1000 US$; just under 30 percent when extended to 1500 US$; and below 25 percent up to 3000 US$ per capita per year. For most countries with a GDP per capita of more than 3000 US$ per year, donor funding makes up a very small share of total expenditure—typically less than five percent (with a few exceptions).
In countries where healthcare is principally financed through public funds, out-of-pocket spending is typically low; this is natural, since in these countries there is essentially universal coverage through public insurance (e.g. Cuba, UK, Sweden, France). And by the same logic, out-of-pocket spending is also low in countries where healthcare is largely financed through private funds in the form of private voluntary insurance (e.g. US). It is in countries with low public healthcare spending and low private voluntary insurance that out-of-pocket expenditure is high (e.g. India, Afghanistan, Sudan). This is shown in the visualization from Jamison et al (2013)22; it illustrates how much progress different countries have made in providing ‘prepaid care’ and the extent to which they use public funds (compulsory social insurance or funding from general government revenue) or private voluntary insurance. As it can be seen, there are no countries near the top-left corner while there are several in the bottom-right. The message seems to be that achieving universal coverage requires government or publicly mandated finance – arguably a justification for some of the reforms recently implemented by the Affordable Care Act in the US (see discussion immediately below).
Share of healthcare spending from private voluntary insurance vs. share of healthcare spending from public (or publicly mandated) finance, 2011 – Jamison et al (2013)23
In June 2012 the US introduced the Affordable Care Act (ACA) – a legal reform aiming to improve the accessibility, affordability, and quality of healthcare. In other sections in this entry we provide evidence of some of the underlying issues that motivated a reform to improve healthcare coverage in the US.24
Specifically, in the section on long-term perspective we show that the share of uninsured individuals in the US is large and has remained virtually constant during decades of substantial growth in expenditure; and in the discussion immediately above we noted that achieving universal coverage is likely to require government or publicly mandated finance. Here we want to focus on whether insurance coverage indeed improved after the introduction of the ACA. The visualization shows the percentage of individuals in the US without health insurance for the period 1963-2015. As we can see there are two marked changes in the trends separated by a long period of remarkable stability: there is a sharp drop in the number of uninsured in 1965 with the creation of Medicare and Medicaid, then relatively little change for decades, and then another sharp drop in 2012 with the introduction of the ACA. Disaggregated data shows that those states that decided to expand their Medicaid programs saw larger reductions in their uninsured rates from 2013 to 2015, especially when those states had large uninsured populations to start with (see Obama (2016)25 for further discussion of these figures). While strictly speaking this is only descriptive evidence – we cannot know what would have happened to the trends without the introduction of the ACA –, it seems reasonable to assume that the observed improvements in healthcare coverage are indeed a consequence of the ACA.
At a cross-country level, the strongest predictor of healthcare spending is national income (you can find more about measures of national income in our entry on GDP data). The visualization presents evidence of this relationship. The correlation is striking: countries with a higher per capita income are much more likely to spend a larger share of their income on healthcare. In a seminal paper, Newhouse (1977)26 showed that aggregate income explains almost all of the variance in the level of healthcare expenditure (specifically, Newhouse (1977) showed that among a group of 13 developed countries, GDP per capita explained 92 percent of the variance in per capita health expenditure). Other studies have confirmed that this strong positive relationship remains after accounting for additional factors, such as country-specific demographic characteristics.27
Although in strict sense this result cannot be interpreted causally – since countries differ in many unobservable aspects that relate both to income and healthcare spending –, more sophisticated econometric models dealing with the issue of ‘omitted variables’ seem to confirm that the effect of per capita GDP on expenditure is clearly positive and significant (for a technical discussion of this conclusion see Culyer and Newhouse (2000)).28
As we discussed in the previous section on international flows of global health financing, in some countries an important part of the income that is available to finance healthcare comes from international development assistance. Lu et al. (2010)29 use data from the IHME – the same data we discussed in the previous section – to show that there is a negative correlation between development assistance for health provided to governments, and government health funding from autonomous sources. Interestingly, this ‘crowding-out’ effect depends on the channel: the correlation between international health aid to non-governmental organizations and government health funding is negative.
In the previous section we pointed out that healthcare spending from autonomous sources has increased substantially in many low and middle-income countries over the last couple of decades. The chart shows a recent snapshot of the cross-country correlation between tax revenue and healthcare spending in these countries.
We can see that developing countries with higher tax revenues tend to spend more on healthcare. In fact, in a recent paper, Reeves et al. (2015)30 estimate the relationship between tax revenue and access to healthcare in developing countries, and find that tax revenue is an important statistical determinant of progress towards universal health coverage – and this remains true after controlling for country-specific time-invariant factors (country ‘fixed effects’).
If we interpret these results causally – which as usual requires making strong assumptions–, the implication seems to be that increasing domestic tax revenues contributes importantly to achieving universal health coverage, particularly in countries with low tax bases. Exploring the relationship between health outcomes and different types of taxation, Reeves et al. (2015) further suggest that pro-poor taxes on profits and capital gains may support expanding health coverage without the adverse associations with health outcomes observed for higher consumption taxes.31
Empirical evidence suggests that healthcare spending is not only sensitive to changes in income (as discussed above), but in many instances, also sensitive to changes in prices.
As it turns out, price-sensitivity is so critical in low-income countries, that small costs for important healthcare products make a vast difference in demand.32 The graph, from a policy report produced by the think-tank Poverty Action Lab, summarizes the findings from a number of studies testing the link between demand and small price changes in experimental settings investigated in randomized control trials.
Demand for preventive healthcare products based on price – JPAL (2011)33
In most countries with market economies, the market for healthcare is only one of many markets competing for the same resources; because of this the prices for healthcare services are affected by productivity changes in other markets. Economic theory suggests that, if the productivity of the healthcare industry increases slower than that of other industries (a probable scenario given that healthcare provision is particularly labour-intensive), then prices in the healthcare sector are likely to grow faster than inflation, and expenditure as a share of income is thus likely to grow (this argument is known as Baumol’s ‘cost disease’). The graph from Culyer and Newhouse (2000) shows that in the U.S., over the course of the 20th century the growth in the consumer price index for all goods and services (CPI) was lower than the growth in the medical consumer price index (MCPI). You can read more about the composition of the bundle of goods and services measured in the CPI and their relationship to GDP deflators in our entry on GDP data.
Price inflation in the overall consumer price index (CPI) and the medical consumer price index (MCPI), U.S. 1927-1996 – Culyer and Newhouse (2000)34
Healthcare is one of the most important inputs to produce health; and life expectancy is one of the key measures of a population’s health. The visualization shows the relationship between life expectancy at birth and healthcare expenditure per capita. This chart shows the level of both measures in the first and last year for which data is available (1995 and 2014 respectively). The arrows connect these two observations, thereby showing the change over time for all countries in the world. As it can be seen, countries with higher expenditure on healthcare per person tend to have a higher life expectancy. And looking at the change over time, we see that as countries spend more money on health, life expectancy of the population increases.
Notice that the relationship in this chart follows a pattern of ‘diminishing returns’: the additional increase in life expectancy associated with an increase in healthcare expenditure decreases as expenditure increases. This means the proportional highest gains are achieved in countries with low baseline levels of spending. This pattern is similar to that observed between life expectancy and per capita income.
The countries are color-coded by world region, as per the inserted legends. Many of the African countries (in purple) achieved remarkable progress over the last 2 decades: health spending often increased substantially and life expectancy in many African countries increased by more than 10 years. The most extreme case is Rwanda, where life expectancy has increased from 32 to 64 years since 1995, a key additional reason is that this was just one year after the brutal genocide in the country. The graph also shows that the African countries that suffered the most under the HIV/AIDS epidemic — Lesotho, Eswatini, and South Africa — experienced a decline of life expectancy from which they have not yet recovered.
The two most populous countries of the world – India and China – are emphasized by larger arrows. It is interesting to see that in 1995 China achieved good health outcomes at comparatively low levels of health spending.
The association between health spending and increasing life expectancy also holds for rich countries in Europe, Asia, and North America in the upper right corner of the chart. The US is an outlier that achieves only a comparatively short life expectancy considering the fact that the country has by far the highest health expenditure of any country in the world (we will get back to this aspect further below).
The visualization presents the relationship between child mortality – measured as the share of children dying before their fifth birthday – and healthcare expenditure per capita. Global data on health expenditure per capita is available since 1995 and in this chart we show the level of both measures in the first and last year for which data is available. The arrows connect these two observations, thereby showing the change over time for all countries in the world. We can see that child mortality is declining as more money is spent on health.
Focusing on change over time we can see a particularly striking fact: while there is a huge inequality in levels – child mortality in the best-off countries is almost 100-times lower than in the worst – inequality in trends is surprisingly low. Specifically, if you look at the paths over time it is surprising how little heterogeneity there is between very different countries in the world. No matter whether it is a rich country in Europe or a much poorer country in Africa, the proportional decline in child mortality associated with a proportional increase in health expenditure is remarkably similar.
The visualization also shows the very high global inequality in health spending per capita that is still prevalent today. In the Central African Republic only int.-$25 spent per capita while on the other end of the distribution, in the US, int.-$9,403 are spent. The ratio between the two countries is 376; on average Americans spent more on health per day than a person in the Central African Republic spends in an entire year. This is a very large gap, considering that International-$ are adjusted for price differences between countries – if price differences were not taken into account, and the spending would have been expressed in US-$ by simply using the exchange rate between the different currencies, the difference would be even larger.
The results here are similar to the strong correlation between healthcare expenditure and life expectancy that we noted in the previous section.
The above-mentioned cross-sectional relationships cannot be interpreted causally because countries differ in a number of ways that simultaneously affect health outcomes and health expenditure. Income is one of them. But we can get a step closer by concentrating on countries with similar income per capita, and looking at changes across time for each country, which eliminates the potential confounding effect of country-specific time-invariant factors.
The graph visualizes the relationship between life expectancy and health expenditure, for a number of OECD countries since 1970. Two points are worth mentioning. Firstly, all countries in this graph have followed an upward trajectory (life expectancy increased as health expenditure increased), but the U.S. stands out as an exception following a much flatter trajectory; gains in life expectancy from additional health spending in the U.S. were much smaller than in the other high-income countries, particularly since the mid-1980s. And secondly, the gains for all countries (except for the U.S.) were not diminishing, as in the previous graph. This suggests that there are many other factors affecting life expectancy, that are not determined by healthcare spending. Indeed, as we have pointed out before, healthcare is just one of many inputs to produce health.
The ever-elusive question of causality is not fully addressed in the analysis above (the issue of time varying unobservables and potential simultaneity remains), but the presented visualizations are nevertheless indicative of a clear and strong relationship; the health returns to healthcare investment suggested by the figures above are substantial, particularly for low-income countries. Indeed, these returns coupled with the returns estimated for conditional-cash-transfer programs in low and middle income countries,35 suggest that with current world-wide resources – suitably targeted – it could be feasible to drastically reduce world inequality in health outcomes and achieve the so-called “grand convergence”.36
The main source of data on international healthcare expenditure is the World Health Organisation (WHO), more specifically the global health expenditure database. This is the same data published by the World Bank (World Development Indicators) and Gapminder. It is also the source of the health expenditure tables in the World Health Statistics Report and the WHO Global Health Observatory; and it is used as an input to the Development Assistance for Health Database from the IHME.37
WHO data is produced mainly from national reports, but it also relies on reports from international organizations. The health statistics from the Organisation for Economic Co-operation and Development (OECD), specifically the Health Expenditure and Financing Dataset, are one of the inputs used by the WHO. Other important inputs from international organizations are the EUROSTAT database and the United Nations national accounts statistics. Since the WHO relies on a number of sources, its estimates do not always coincide with e.g. national statistics or OECD reports.
Another, related but different source of healthcare expenditure data, is the International Food Policy Research Institute (IFPRI), which publishes the Statistics of Public Expenditure for Economic Development (SPEED). This source relies primarily on data from the International Monetary Fund (IMF).
In the WHO’s global health expenditure database (GHED), total expenditure on health is calculated as the sum of public health expenditure (also referred to as general government expenditure) and private health expenditure. The former is calculated by adding up all outlays paid by government entities (ministries, parastatal organizations or social security agencies), regardless of the source (so includes in principle any donor funding passing through them). The latter is calculated by adding up outlays for health by private entities (households, commercial health insurance, non-profit institutions). The WHO produces an Indicator Codebook that provides detailed information about the definitions of all of the variables in their dataset.
The GHED has been maintained for the past ten years with the intent of providing comparable expenditure figures across countries and time, and its estimates rely on national health accounts based on reports from ministries, central banks, national statistics offices and international agencies such as the OECD. Since not all countries have (or update) national health accounts, the GHED’s estimates often require imputation for missing values. Lu et al. (2010)38 provide an excellent account of the quality of the resulting estimates. Their conclusion is that the WHO’s adjustment process requires substantial imputations for missing data; and that the distinction between actual reported data and imputed estimates is not clearly stated in their published tables. After reviewing the methodology used by the WHO, the authors concluded that the imputation methods are not standardised, and that the imputations are often based on the assumption that the ratio of government health spending to general government spending remains constant with time.
As mentioned before, the IMF also tracks public spending on health, and is one of the underlying sources used by the WHO for their GHED estimates. Lu et al. (2000) compare IMF reports with WHO estimates and conclude that the correlation between them is only 0.65. They argue the discrepancies arise because IMF values are more likely to represent data from ministries of finance than from ministries of health.
Apart from the above-mentioned issues, international cross-country comparisons of healthcare financing estimates are also problematic due to lack of consistency in classification of expenditures, especially on the borderline of health services. Even across OECD countries, where there is a framework of harmonised accounting, there are issues of completeness and comparability, particularly across time.39
An example of this problem is the classification of care for disabilities. Culver and Newhouse (2000) mention that expenditure on intellectual developmental disabilities is not included in the healthcare figures from Denmark and Sweden after 1985, but it is included in the figures from Finland, Iceland and Norway.
As it has been mentioned, the earliest data on financing of healthcare dates back to the late 19th century, when many European countries began officially establishing healthcare systems through legislative acts. The main sources here are academic publications.
- Data Source: Lindert, Peter H. “The rise of social spending, 1880-1930.” Explorations in Economic History 31, no. 1 (1994): 1-37.
- Description of available measures: Public Health Expenditure as percent of GDP
- Time span: 1880-1930
- Geographical coverage: Selection of high-income countries
- Data Source: Flora, Peter et al. 1983. State, Economy and Society in Western Europe, 1815-1975. Frankfurt: Campus Verlag
- Description of available measures: Central government expenditure by sectors, as percent of GDP and as percent of total expenditure.
- Time span: 1815-1975
- Geographical coverage: Western Europe
- Link: Available online from http://gpih.ucdavis.edu/Government.htm/
- Data Source: Tanzi, Vito, and Ludger Schuknecht. Public spending in the 20th century: A global perspective. Cambridge University Press, 2000.
- Description of available measures:
- Public Health Expenditure as percent of GDP
- Health Insurance Coverage as percent of labour force
- Time span: 1910-1994
- Geographical coverage: Selection of high-income countries
Country-specific statistics are another important source of long-run data on healthcare spending. Two references here are the U.S. Bureau of the Census and the Office of Health Economics in the U.K.
- Data Source: (a) Historical Statistics of the United States Colonial Times to 1970 (1929-1970); and (b) US Census Statistical Abstract 1990 (1970-1990). Both published by the US Bureau of the Census
- Description of available measures: Total healthcare expenditure, disaggregated by private and public spending, with further details on specific medical services (figures expressed in current prices, as well as shares of GDP)
- Time span: 1929-1990
- Geographical coverage: U.S.
- Data Source: Office of Health Economics, based on Annual Abstract of Statistics, and Government’s Expenditure Plans
- Description of available measures: Total healthcare expenditure, disaggregated by private and public spending, with further details on specific medical services and NHS expenditure (figures expressed in current and inflation-adjusted prices, as well as shares of GDP)
- Time span: 1950-2011
- Geographical coverage: U.K.
The most common source of up-to-date cross-country healthcare expenditure is the Global Health Expenditure Database from the WHO. This is the same data published by the World Bank (World Development Indicators) and Gap Minder.
- Data Source: WHO based on reports from ministries, central banks, national statistics offices and international agencies
- Description of available measures:
- External resources for health as a percentage of total expenditure on health
- General government expenditure on health as a percentage of total expenditure on health
- General government expenditure on health as a percentage of total government expenditure
- Out-of-pocket expenditure as a percentage of private expenditure on health
- Out-of-pocket expenditure as a percentage of total expenditure on health
- Per capita government expenditure on health (PPP int. $)
- Per capita government expenditure on health at average exchange rate (US$)
- Per capita total expenditure on health (PPP int. $)
- Per capita total expenditure on health at average exchange rate (US$)
- Private expenditure on health as a percentage of total expenditure on health
- Private prepaid plans as a percentage of private expenditure on health
- Social security expenditure on health as a percentage of general government expenditure on health • Total expenditure on health as a percentage of gross domestic product
- Time span: 1994-2013
- Geographical coverage: Global by Country
- Link: http://apps.who.int/nha/database
Another important source is the OECD’s Health Expenditure and Financing Dataset. This feeds, in part, the WHO’s Global Health Expenditure Database described above.
- Data Source: OECD based on reports from member countries as per national accounts framework
- Description of available measures:
- Total and current expenditure on health Expenditure on personal health care
- Expenditure on collective health care
- Expenditure on services not allocated by function
- Additional health expenditure aggregates
- Expenditure on health-related functions
- Current health expenditure by provider
- Expenditure by age and gender
- Time span: 1960-2014, though significant missing values prior 1990
- Geographical coverage: OECD countries
- Link: http://stats.oecd.org
Yet another relevant source of internationally comparable expenditure statistics is IFPRI’s Statistics of Public Expenditure for Economic Development. This dataset relies mainly on IMF statistics, which are also an input to the WHO’s Global Health Expenditure Database described above
- Data Source: IFPRI, from multiple dats sources, but mainly IMF statistics
- Description of available measures:
- health expenditure in 2005 $ppp
- health expenditure in 2005 US$
- percentage of health expenditure in total gdp
- per capita health expenditure in 2005 $ppp
- percentage of health expendtiure in total expenditure
- Time span: 1980-2012
- Geographical coverage: 67 countries across all continents
- Link: https://www.ifpri.org/publication/statistics-public-expenditures-economic-development-speed
The Global Health Data Exchange, created and supported by IHME, catalogs a vast amount of information pertaining to the healthcare sector.