Public spending enables governments to produce and purchase goods and services, in order to fulfil their objectives – such as the provision of public goods or the redistribution of resources. In this entry we study public spending through the lens of aggregate cross-country data on government expenditures. We begin with an analysis of historical trends, and then move on to analyze recent developments in public spending patterns around the world.
The available long-run data shows that the role and size of governments around the world has changed drastically in the last couple of centuries. In early-industrilized countries, specifically, the historical data shows that public spending increased remarkably in the 20th century, as governments started spending more resources on social protection, education and healthcare.
Recent data on public spending reveals substantial cross-country heterogeneity. Relative to low-income countries, government expenditure in high-income countries tends to be much larger (both in per capita terms, and as share of GDP), and it also tends to be more focused on social protection.
Recent data on public spending also shows that governments around the world often rely on the private sector to produce and manage goods and services. And public-private partnerships (PPP), in particular, have become an increasingly popular mechanism for governments to finance, design, build and operate infrastructure projects. In the period 2005-2010 alone, the total value of PPP projects in low and middle-income countries more than doubled.
The visualization shows the evolution of government expenditure as a share of national income, for a selection of countries over the last century. The source of the data is Mauro et al. (2015).1
The long-run series in this dataset cover mainly, but not exclusively OECD countries. Non-OECD countries with available long-run data include Russia, India, Argentina, Brazil, Peru and Colombia.
The above-mentioned long-run series are complemented in this dataset by comparable recent estimates for most countries in the world. You can plot other countries in this visualization by selecting ‘Add Country’; but bear in mind that the series for most non-OECD countries are much shorter.
If we focus on early-industrialized countries, we can see that there are four broad periods in this chart. In the first period, until the First World War, spending was generally low. In the US, for example, total government expenditure accounted for less than 2% of national income until 1916. These low levels of public spending were just enough for governments to be concerned with basic functions, such as maintaining order and enforcing property rights.
In the second period, between 1915-1945, public spending was generally volatile, particularly for countries that were more heavily involved in the First and Second World Wars. Government expenditures as a share of national output went sharply up and down in these countries, mainly because of changes in defense spending and national incomes. In the US, public spending as a share of GDP was 10.5% in 1941, then went up to 44.1% in 1945, and then went back down to 12.2% in 1948.
In the third period, between 1945-1980, public spending grew particularly fast. As we show in more detail later, this was the result of growth in social spending; and was largely made possible by historical increases in government revenues over the same period.
Since 1980 the growth of government expenditure has been slowing down in early-industrialized countries – and in some cases, it has gone down in relative terms. However, in spite of differences in levels, in all these countries public spending as a share of GDP is higher today than before the Second World War.
Although the increase in public spending has not been equal in all countries, it is still remarkable that growth has been a general phenomenon, despite large underlying institutional differences.
The map uses the same data, for all countries, to show global patterns. You can switch between the ‘Map’ and ‘Chart’ views by selecting the corresponding tabs in the interactive visualziation. By using the slider at the bottom you can get a sense of the long-run trend in global government expenditures.
At the end of the 19th century European countries spent less than 10% of GDP via the government. In the 21st century this figure exceeds 50% in many European countries. The increase in absolute terms – rather than the shown relative terms – is much larger since the level of GDP per capita increased very substantially over this period.
The visualization above shows that government spending in early-industrialised countries grew substantially in the 20th century. The visualization shows that this was the result of growth specifically in social spending.
The visualization maps recent estimates of central government expenditure, as share of national incomes, across the world. The data is published as part of the World Development Indicators and comes from the IMF.
The most striking feature in the chart is the degree of heterogeneity between world regions. Central governments in high-income countries – particularly those in Europe – tend to control a much larger share of national production than governments in low-income countries.
In countries such as France, central government spending accounts for almost 50% of all national output. In Nigeria the corresponding figure is close to 6%.
Although there are many missing observations, the data suggests that the first decade of the 21st century was characterized by a broad positive trend. You can explore more detailed country-specific time trends by selecting the ‘Chart’ tab, located on top of the map.
These estimates have to be interpreted with caution, since central government expenditure provides a somewhat distorted picture of total public spending, particularly in federal countries with large sub-national governments. As we discuss in the data quality section, it is unfortunately hard to find reliable cross-country data on expenditure below the central level.
The visualization shows total public expenditure per capita, across all levels of government, for OECD countries. To allow for cross-country comparability, these estimates are expressed in current PPP US dollars. Some countries that are not members of the OECD, but that do report comparable statistics, are also included in this graph.
As we can see, cross-country differences are persistently large. In India, the government spends about 1,800 US dollars per head (PPP) in one year; while in countries such as Norway, the latest corresponding figure is just over 30,000 US dollars (PPP).
The visualizations above show that governments around the world differ considerably in size, even after controlling for underlying differences in economic activity and population. Here we show that, as one would expect, governments also differ substantially in terms of how they prioritise expenditures.
The visualization shows the share of government expenditure that is specifically allocated to education.
As we can see, there are large and persistent differences, even within developing countries. For example, in 2011 education accounted for about 8% of government spending in the Central African Republic, while it accounted for about 30% in Ghana.
We have already pointed out that governments in high-income country spend more resources than governments in low-income countries, both in per capita terms, and as share of their national incomes. Here we focus on the social spending component of government expenses, and show that high-income countries also have higher levels of social spending, particularly in the form of transfers.
The visualization, from Bastagli et al (2012)2, shows stacked social expenditure figures for different country groups. The different color segments in the bars reflect different components of social spending. The sizes reflect expenditure as share of GDP.
As we can see, the group of ‘advanced economies’ dedicates a much larger share of national income specifically to social transfers. This contrasts with the figures from sub-Saharan Africa, where social spending is much lower across the board, and where transfers play a less important role.
The visualization above emphasizes regional differences, but masks a great deal of heterogeneity within regions. The visualization shows social protection expenditures as a share of total general government expenditures, across different OECD countries. As we can see, in countries such as Denmark and Luxembourg, more than 40% of total government spending goes to social protection. In South Korea and the US, the corresponding figures are close to 20%.
Levels and composition of public social spending (% GDP), World Regions- Figure 6 in Bastagli et al (2012)3
In the chart we see the allocation of public spending (given as the percentage of a country’s GDP) across a range of ‘social spending’ branches. Note that you can switch between OECD countries, and can switch from ‘absolute’ to ‘relative’ figures.
Although there are some small cross-country variations in the way social expenditure is distributed, the three priorities are predominantly the same across the OECD. Old age expenditure (in the form of pensions and elderly care) typically receives the largest allocation of social spending, followed by health, with either family or incapacity-related benefits typically coming in third.
On average, OECD countries spend 7-8 per cent of GDP on old age care, 6 per cent on health, and 2 per cent on both family and incapacity-related benefit. And the relative importance of these branches has remained largely constant since 1980.
The visualization shows the share of central government expenditure that goes to the compensation of government employees. Compensation of employees includes all salaries and benefits (both in cash and in kind).
As we can see, the salaries of public servants and other government employees are an important component of public spending in most countries. Yet differences between countries are very large.
Throughout Europe, the share of government spending that is devoted to the compensation of government employees ranges between 5% and 15%. In contrast, throughout most of Africa the available figures range between 30% and 50%.
Governments around the world often rely on the private sector to produce and manage goods and services. The process through which governments purchase works, goods and services from companies, which they have selected for this purpose, is often referred to as ‘public procurement’.
The two visualizations show the importance of public procurement in OECD countries (and partner countries providing comparable data). The first chart shows the value of total general government procurement as percentage of GDP, while the second chart shows the relative weight of procurement within total expenditure (i.e. general government procurement as a percentage of total government expenditures).
As we can see, public sector purchases from the private sector are significant in many high-income countries. In the Netherlands, almost 45% of total government expenditure is channeled through procurement. This corresponds to about 20% of the national income of the Netherlands. In Greece, the relative weight of procurement in the domestic budget is much smaller (about 20% of total expenditure), yet size remains significant for the economy (about 10% of GDP).
Public procurement comprises many different forms of purchases. Public procurement includes, for example, tendering and contracting in order to build large infrastructural projects. However, public procurement goes beyond infrastructure. It also includes, for example, purchases of routine office supplies.
Generally speaking, the part of public procurement that does not fall within the category of gross fixed capital formation (e.g. building new roads), is referred to as ‘outsourcing’, or ‘contracting-out’. This form of procurement often relies on short-term contracts.
According to the definitions used by the OECD, outsourcing includes both intermediate goods used by governments (such as procurement of information technology services), or the outsourcing of final goods and services financed by governments (such as social transfers in kind via market producers paid for by governments).
The visualization shows total expenditures on general government outsourcing (accounting for both intermediate and final goods), as a share of GDP. The data is from 2016, and is available only for a selection of OECD member countries (plus a few other partner countries reporting data to the OECD under the same methodology).
As we can see, governments in many high-income countries spend substantial resources via outsourcing. In the Netherlands, where general procurement is large, outsourcing purchases account for almost 17% of national income.
Public procurement strategies available to governments are varied. Governments may choose to take responsibility for financing, designing, building and operating infrastructure projects – and they simply outsource specific elements. Or they may choose to pursue a public-private partnership, where private actors directly take responsibility for all these aspects, from financing to operation.
The term ‘private finance initiative’ is often used to denote a public procurement strategy, whereby governments choose a private firm (or consortium) to construct and operate – and sometimes also finance – public infrastructure. These initiatives typically take the form of long-term contracts. The term ‘public-private partnerships’ is often used to denote those private finance initiatives where the public sector retains an important participation. More information about terms and classification methodologies can be found in the resources provided by the World Bank’s Private Participation in Infrastructure Database (PPID).
The chart, from the World Bank’s PPID, shows the evolution of public-private partnerships in infrastructure, aggregating projects across 139 low and middle-income countries. The blue series shows the total value of projects in US dollars (scale in the left vertical axis), while the orange series shows the total number of projects (scale in the right vertical axis).
As we can see, the last two decades have seen a marked increase in public-private partnerships in low and middle-income countries. In the World Bank’s PPID Visualization Dashboard you can explore the data in more detail. The estimates by sectors and world regions suggests that electricity and roads, specifically in South Asia and Latin America, have been the key drivers of these aggregate trends.
Total investment (billions of US dollars) and number of PPP projects in low and middle income countries, 1990-2015 – World Bank Private Participation in Infrastructure Database
We have already pointed out that government expenditure as a share of national income is higher in richer countries. The visualization provides further evidence of the extent of this correlation.
The vertical axis measures GDP per capita (after accounting for differences in purchasing power across countries), while the horizontal axis measures governments spending as share of GDP. The vertical axis is expressed by default in a logarithmic scale, so that the correlation is easier to appreciate – you can change to a linear scale by clicking the ‘Log’ button.
We can see that there is a strong positive correlation: high-income countries tend to have larger government expenditures as a share of their GDP. And this is also true within world regions (represented here with different colors).
This correlation reflects the fact that high-income countries tend to have more capacity to extract revenues, which in turn is due to their capacity to implement efficient tax collection systems. In our entry on Taxation we discuss the drivers of tax revenues in detail.
The visualization shows the reduction of inequality that different OECD countries achieve through taxes and transfers.
The estimates correspond to the percentage point reduction in inequality, as measured by changes in the Gini coefficients of income, before and after taxes and transfers. The source for the data is the 2016 OECD Inequality Update as part of the OECD Income Distribution Database (IDD) November 2016 Release: Income inequality remains high in the face of weak recovery. The IDD provides further details regarding how these estimates are constructed. In a nutshell, income ‘before taxes’ corresponds to what is usually known as market income (wages and salaries, self-employment income, capital and property income); while income after taxes and transfers corresponds to disposable income (market income, plus social security, cash transfers and private transfers, minus income taxes).
The data shows that across the 35 countries covered, taxes and transfers lower income inequality by around one-third on average (equivalent to around 0.15 Gini points). Yet cross-country differences are substantial, with declines ranging from about 40% in Denmark and Ireland, to about 8% in South Korea. The US – a country with high baseline levels of inequality – achieves a reduction of around 17%, which is a little over half of the OECD average.
Generally speaking, countries that achieve the largest redistribution through taxes and transfers tend to be those with the lowest after-tax inequality.
The main source of cross-country data on aggregate government expenditures is the IMF. According to the IMF, government expenditure is calculated as the sum of all cash payments for operating activities of the government in providing goods and services, including compensation of employees (such as wages and salaries), interest and subsidies, grants, social benefits, and other expenses such as rents and dividends.
The IMF, through its Government Finance Statistics Manuals and Guides, recommends an “accrual accounting method, focusing on all economic events affecting assets, liabilities, revenues, and expenses, not just those represented by cash transactions”. By construction, the recommended method accounts for changes in stocks, which means that “stock data at the end of an accounting period equal stock data at the beginning of the period plus flows over the period.” You can learn more about the definitions and accounting methods directly from the IMF Government Finance Statistics Manuals and Guides.
The most important limitation with the estimates produced by the IMF is that, despite their efforts to standardize data collection, many countries report misclassified, incomplete, and untimely statistics.
The lack of consistent data on local government expenditures often makes cross-country comparisons difficult. Because of this, many studies rely on central government estimates, even though these estimates provide an incomplete picture, especially in federal countries. Additionally, the documentation for the World Development Indicator on Government Expenditure – which uses IMF estimates – notes: “For most countries central government finance data have been consolidated into one account, but for others only budgetary central government accounts are available. Because budgetary accounts may not include all central government units (such as social security funds), they usually provide an incomplete picture.”
The chart provides a comparison of two cross-country measures of government expenditure. The first measure, in the horizontal axis, corresponds to the World Development Indicators, and as mentioned above, corresponds mainly to central government spending. The second measure, in the vertical axis, corresponds to the depurated estimates from Mauro et al. (2015), where the authors tried to incorporate expenditure data across all government levels.4 Note that interest payments on debt have been included in both measures of government expenditure.
As we can see, while the two measures are correlated, they are still substantially different. A big part of the difference between these two measures can be attributed to the fact that one of them accounts only for central government expenditures – indeed, most countries lie above a line with slope one, which suggests that local government expenditure is not negligible. There remains a few countries where there is significant mismatch between total and central government spending- in the case of Costa Rica and Afghanistan, central spending appears to exceed total government spending. In both countries, there appears to be notable inconsistencies both in terms of budget allocation and actual funds distribution between governmental and sub-national entities, and the approval process for spending autonomy by decentralised institutions.5,6
Misclassification and poor standardised within country statistics therefore remains a challenge.
The above-mentioned limitations are substantially less critical for estimates reported to the OECD by member countries, mainly because they tend to apply more rigorously the agreed international conventions and accounting methods – including a consistent accounting of expenditures across sectors and levels of government. However, for countries that are not members or partners of the OECD, the data limitations are severe, and cross-country comparisons over time have to be interpreted with caution.
In addition to the IMF and the OECD, the WHO and UNESCO also collect and report data on healthcare and education government expenditures. You can read about these sources, including their relationship and limitations, in our entries on Financing Healthcare and Financing Education.
In recent years, the International Food Policy Research Institute (IFPRI) started producing a depurated database of cross-country data on government spending, called SPEED (Statistics on Public Expenditures for Economic Development database).
The 2015 version of IFPRI’s SPEED database contains information from 10 public expenditure sectors in 147 countries from 1980 to 2012. To produce this database “IFPRI researchers have compiled data from multiple sources, including the International Monetary Fund, the World Bank, and national governments, and conducted extensive data checks and adjustments to ensure consistent spending measurements over time that are free of exchange-rate fluctuations and currency denomination changes.”
Unfortunately, despite IFPRI’s best efforts, we have reasons to believe that there are substantial remaining problems with the estimates published in the SPEED database. For example, the estimates of government expenditure as a share of GDP in Zimbabwe and Palestine are in the thousands for several years. In such complicated cases it would be better not to report estimates – even more so considering that the SPEED database reports regional and global averages that are not country-weighted.
- 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: Social spending as percent of GDP
- Time span: 1880-1930
- Geographical coverage: Selection of high-income countries
Flora et al. (1983)
- 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 (total and 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/
Tanzi and Schuknecht (2000)
- 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 Expenditure (total and by sectors) as percent of GDP
- Time span: 1910-1994
- Geographical coverage: Selection of high-income countries
Mauro et al. (2015)
- Data Source: IMF Fiscal Affairs Departmental Data. Based on Mauro, P., Romeu, R., Binder, A., & Zaman, A. (2015). A modern history of fiscal prudence and profligacy. Journal of Monetary Economics, 76, 55-70.
- Description of available measures: Total government expenditure, excluding interest payments, as share of GDP
- Time span: 1880-2011
- Geographical coverage: Global by country
- Link: http://www.imf.org/external/datamapper/index.php
Roine, Vlachos, Waldenström (2009)
- Data Source: Roine, J., Vlachos, J., & Waldenström, D. (2007). What Determines Top Income Shares? Evidence from the Twentieth Century (No. 2007: 17). Stockholm University, Department of Economics.
- Description of available measures: Central government expenditure as share of GDP
- Time span: 1900-2005
- Geographical coverage: Selected countries (mainly high-income)
- Link: http://www.ifn.se/Wfiles/wp/wp721.pdf
IMF via World Development Indicators
- Data Source: IMF Government Finance Statistics Yearbook and data files
- Description of available measures: Central government expenditure as share of GDP (certain types of transactions, such as employee compensations, are provided separately, both as share of GDP and as share of total expenditure)
- Time span: Longest series, 1972-2014, but most countries have much shorter series
- Geographical coverage: Global by country
- Link: http://databank.worldbank.org/data/home.aspx
IFPRI – SPEED
- Data Source: IFPRI, from multiple data sources, but mainly IMF statistics
- Description of available measures: Central government expenditure (total and by sectors) as share of GDP
- Time span: 1980-2012
- Geographical coverage: Global by country
- Link: https://www.ifpri.org/publication/statistics-public-expenditures-economic-development-speed
OECD – Government at a Glance
- Data Source: OECD
- Description of available measures: Government expenditures (by level of government, type of transaction and sector). Indicators are available as share of GDP (some are also available in per capita PPP USD).
- Time span: Longest series are, 1972-2014 (some countries have shorter series)
- Geographical coverage: OECD member countries (plus several partner countries that report under the same methodology)
- Link: http://stats.oecd.org/
Armingeon et al. (2016) – Comparative Political Dataset
- Data Source: Armingeon, Klaus, Christian Isler, Laura Knöpfel, David Weisstanner and Sarah Engler. 2016.
Comparative Political Data Set 1960-2014. Bern: Institute of Political Science, University of Berne.
- Description of available measures: Total outlays (disbursements) of general government as a percentage of GDP
- Time span: 1960-2011
- Geographical coverage: Selected OECD countries
- Link: http://www.cpds-data.org/index.php/data#CPDS
Private Participation in Infrastructure Database
- Data Source: World Bank Private Participation in Infrastructure Database
- Description of available measures: Value and number of public-private partnership projects (by country, sector, and type of partnership)
- Time span: 1990-2015
- Geographical coverage: low- and middle-income countries (World Bank classification)
- Link: http://ppi.worldbank.org/data