The Gross Domestic Product (GDP) of an economy is a measure of total production. More precisely, it is the monetary value of all goods and services produced within a country or region in a specific time period. While the definition of GDP is straightforward, accurately measuring it is a surprisingly difficult undertaking. Moreover, any attempts to make comparisons over time and across borders are complicated by price, quality and currency differences. This article covers the basics of GDP data and highlights many of the pitfalls associated with intertemporal and spatial comparisons.
From the long-term perspective of social history, we know that economic prosperity and lasting economic growth is a very recent achievement for humanity. In this section we will look at this more recent time and will also study the inequality between different regions – both in respect to the unequal levels of prosperity today and the unequal economic starting points for leaving the poverty of the pre-growth past.
Economic prosperity is measured as via growth domestic product (GDP) per capita, the the value of all goods and services produced by a country in one year divided by the country’s population. Economic growth is the measure of the change of GDP from one year to the next. This entry shows that the current experience of economic growth is an absolute exception in the very long-run perspective of social history.
# Empirical View
# From poverty to prosperity: The UK over the long run
The UK is particularly interesting as it was the first economy that achieved sustained economic growth and thereby previously unimaginable prosperity for the majority of the population.
# Output per capita of the UK economy
The chart below shows the reconstructed GDP per capita in England and the UK over the last 7 centuries.
Incomes remained almost unchanged over a period of several centuries when compared to the increase in incomes over the last 2 centuries. Life too changed remarkably little. What people used as shelter, food, clothing, energy supply, their light source stayed very similar for a very long time. Almost all that ordinary people used and consumed in the 17th century would have been very familiar to people living a thousand or even a couple of thousand years earlier.
# Output of the UK economy
# The economy before economic growth: The Malthusian trap
# The pre-growth economy was a zero-sum-game: Living standards were determined by the size of the population
In the previous chart we saw that it was only after 1650 that living standards in the UK did start to increase for a sustained period. Before the modern era of economic growth the economy worked very differently. Not technological progress, but the size of the population determined the standards of living.
If you go back to the chart of GDP per capita in the England you see that early in the 14th century there was a substantial spike in the level of incomes. Incomes increased by around a third in a period of just a few years. This is the effect that the plague – the Black Death – had on the incomes of the English. The plague killed almost half(!) of the English population. The population declined from 8 million to 4.3 million in the three years after 1348. We even see it in the chart for the world population.
But those that survived the epidemic were materially much better off afterwards. The economy was a brutal zero-sum game and the death of your neighbour was to the benefit for those that did survive.
This happened primarily because farmers now achieved an higher output. While farmers before the plague had to use agricultural land that was less suited for farming, after the population decline they could farm on the most productive areas of the island.
In the very long time in which humanity was trapped in the Malthusian economy it was births and deaths that determined incomes. More births, lower incomes. More deaths, higher incomes.
We see this coupling of income and population in the chart below that plots the size of the population (on the x-axis) against the total output of the English economy (top panel) and against the income per person (bottom panel). Looking at the bottom panel we see the spike of incomes that was associated with the killing of half of the population in the Black Death. After this the population and the income per person stagnate until around 1500. In the following period we see the economy growing – total GDP increases by more than 280% from 1500 to 1650 – but this increase in output is not associated with an increase in income per person, but only an increase of the total population of the UK.
It is only after 1650 that the English economy breaks out of the Malthusian Trap and that incomes are not determined by the size of the population anymore. For the period after 1650 we see that both the population and the income per person are growing. The economy is not a zero-sum game anymore; economic growth made it a positive-sum game.
When Malthus raised the concerns about population growth in 17981 he was wrong about his time and the future, but he was indeed right in his diagnosis of the dynamics of his past. The world before Malthus was Malthusian and population increases were associated with declining nutrition, declining health, and declining incomes. The world after Malthus became increasingly less Malthusian. What Malthus did not foresee was that the increasing output of the economy will decouple from the change of the population so that the output available for all will increase over a long period. This decoupling of income and population is shown in the chart.
# Technological change in the pre-growth economy
Technological innovation that increases productivity is the key to increased prosperity. But there were technological breakthroughs before the 17th century. Windmills, irrigation technology, and also non-technical novelties especially the new crops from the New World. Why did these not lead to sustained economic growth?
What happened as a consequence of these innovations were indeed increases in productivity, and the output increases led to increased prosperity. But only for a short time. Improvements in technology had a different effect in the Malthusian pre-growth economy. They raised living standards only temporarily and instead raised the size of the population permanently. The economic historian Gregory Clark sums it up crisply: “In the preindustrial world, sporadic technological advance produced people, not wealth.”2
Technological improvements lead to larger, but not richer populations. If this analysis of the pre-growth economy is true than we would expect to see a positive correlation between productivity and the density of the population.
Ashraf and Galor (2011)3 studied the Malthusian economy theoretically and empirically in a paper published in the American Economic Review. The chart below is taken from their publication and confirms the theoretical prediction for the pre-growth economies in the year 1500.
# Rising output by industry
The visualization below shows the rising output of the economy by industry. Each time-series is indexed to the year 1700 so that the focus here is on the change over time as all changes are relative to that year.
The rising output of key industrial and service sectors is shown here.
# Globally Over the last two millennia until today
# The total output of the world economy over the last two thousand years
Data on economic growth is now routinely published by statistical offices, but researchers have had to reconstruct accounts of the economic productivity for the past. These reconstructions are arguably very uncertain. Nevertheless it is absolutely clear that compared to the prolonged growth of economic productivity in the last centuries, the productivity was always very low before: The changes of prosperity over time are much greater than the uncertainty of the exact values in any particular moment in time.
Here I have included one reconstruction of global GDP over the very long run: the last two millennia.
I have used recent data from the World Bank for the period from 1990 onward and for the historical estimates before 1990 I rely on the historical estimates constructed by the economic historian Angus Maddison.4
# Economic output per person around the world over the last two thousand years
There are many reconstructions of GDP per capita over the last centuries; here I will focus on the reconstructions by the British economist Angus Maddison who was working in Groningen (Netherlands) and where, after his death in 2010, younger colleagues are advancing his work in the ‘Maddison Project’.5
As in the previous chart I have used recent data from the World Bank and extended it backwards relying on the estimates of the Maddison project.
What we learn from this chart is that on average the people of the past were many times poorer than we are today. In 1820 the global GDP per capita is estimated to be around 1,230 international-$ per year and this is already after some world regions have achieved economic growth. For all the hundreds, and really thousands, of years before 1820, the average GDP per capita was lower than 1,230 international-$.
Prosperity is a very recent achievement that distinguishes the last 10 or 20 generations from all of their ancestors. In 2015, the average GDP per capita was 14,700 international-$ – more than 10 times the average of the past.
It is often the case that progress creates inequality between regions because it is not happening equally as fast everywhere. If we compare the economic prosperity of every region in 2003 with any earlier time we see that every single region is richer than ever before in its history. Though some regions are more productive than others, every region is doing better than ever before, much better. We deal study this aspect in more detail in our entry on global income inequality.
From the data that we have discussed previously, we know that with respect to economic growth all the action really just happened very recently. It is true that in the pre-growth era some people were very well off – but this was the tiny elite of the tribal leaders, pharaohs, kings and religious leaders. The economic inequality in pre-modern societies was extremely high and the average person was living in conditions that we would call extreme poverty today.
The destitution of the common man only changed with the onset of economic growth. The time when this change happened in various countries is depicted in the following graph. Economic prosperity was only achieved over the last couple of hundred years. In fact, it was mostly achieved over the second half of the last hundred years.
# Economic growth at the technological frontier – growth in the USA
The following chart shows economic growth in the USA adjusted for inflation.
GDP per capita in the USA at the eve of independence was just below $2,000, adjusted for inflation and measured in prices of 2011.
In 2015 – 236 years after independence – GDP per capita has increased more than 26-fold to $52,706. This means that the output per person in one year in the past was the same as the output per person in two weeks today.
It is remarkable how steady economic growth was over this very long period. Over the last two centuries GDP per person in the U.S. economy has grown at around 2 percent per year with only very short deviations from this very steady trend.
# GDP Growth since 1950
# Penn World Table
# World Bank data
# Economic in all countries of the world over the last half century
The following chart plots, for each country, the national income in 1960 against the corresponding national income in 2014. GDP per capita is used to measure national incomes, and figures are expressed in ‘real terms’, which means they are adjusted for inflation.
In this chart, if incomes are stagnant, we should observe countries lining closely along the blue 45° line. Countries in which the income in 2014 is higher than the income in 1960, on the other hand, are above this 45° line. These are all the countries that experienced income growth over these 54 years.
As we can see, some countries such as Madagascar, Chad, Senegal, and Nicaragua stagnated in terms of incomes – they are right on the 45° degree line. And a couple of countries such as Niger and the Democratic Republic of Congo have even experienced negative growth over the reference period. But the large majority of countries, all those above the blue line, have experienced growth.
Those countries that are far above the blue line had the strongest growth. Botswana (38-fold increase), South Korea (30-fold), Romania (15-fold), China (11-fold), and Thailand (18-fold) are some of the countries with the strongest growth over these 54 years.
# Correlates, Determinants, and Consequences
# Productivity is the driver of economic growth
# Productivity – Output over time
# Urbanisation and prosperity
Urbanisation and economic prosperity are strongly correlated as the following visualistion shows. More than 60% of the population lives in cities in all countries with a GDP per capita higher than 40,000 international-$.
# Religiosity and prosperity
A survey asked the question “How important is religion in your life?” and the possible answers were “very important”, “somewhat important”, “not too important” and “not at all important”. The following chart plots the share that answered “very important” against the average prosperity of the population for each country in the survey.
There is a clear correlation between poverty and religiosity. In poor countries the huge majority say that religion is very important in their life: in countries like Uganda, Pakistan, and Indonesia it is the answer of more than 90%. In Ethiopia it is the answer of 98% of the population.
In richer countries the share of the population for whom religion is very important is much lower. In the UK, South Korea, Germany, or Japan it is less than 1 in 5 for whom religion is very important.
The big outlier in this correlation is the USA, a very rich country in which more than 50% answer that religion is very important in their life.
# Retirement becomes possible when people get richer
The visualisation shows the very substantial decline in the labor force participation of men of 65 years and older in the USA since the end of the 19th century.
# Access to financial services
To allow saving and facilitate transactions access to financial services is important. We know that in poorer countries this access is often very limited.
We don’t know much about how this access has changed over time and to understand this change better we have attempted to combine different sources – the result of which is shown in the world map below.
The World Bank Global Financial Inclusion data is available for 2011 and 2014. The data is very scarce on this pre-2011, but World Bank estimates provide an additional single point for countries. This has been represented as a point for 2005, however it’s important to note that this varies between 2000-2005 across countries.
The challenge is that it is not exactly the same measure as the 2011 and 2014 data, but instead a composite measure of access to a bank account and financial services. The World Bank define this composite indicator as measuring “the percentage of the adult population with access to an account with a financial intermediary. The indicator is constructed as follows: for any country with data on access from a household survey, the surveyed percentage is given. For other countries, the percentage is constructed as a function of the estimated number and average size of bank accounts as discussed in Honohan (2007). These numbers are subject to estimation error.”
The use of composite measures is, of course, not ideal. However, we think it should still give a fairly reasonable basis of the early 2000s to use as an earlier estimate and the direction of progress trends.
We have combined this composite measure with 2011 and 2014 data in the following chart.
# Data Quality & Definitions
There are many reasons why we might be interested in the size of an economy, its performance over time, or its relative performance compared to other countries. GDP is the most common way of measuring the size of an economy.
GDP is published in a country’s National Accounts. These statistics comply to protocols laid down in the 1993 version of the Systems of National Accounts, SNA93.
The SNA93 established the ‘production boundary’, a crucial definition of what is and is not included in GDP figures:
- included are all goods and services that are exchanged,
- included are all goods that are not exchanged (for example food produced for home consumption)
- but excluded are services that are not exchanged. Among these excluded services are food preparation, education of children at home, and minor home repairs.
- An important exception to the services that are included is the housing services consumed by owner-occupiers. This service is imputed (imputed rent) and included in the GDP figure.
# GDP per capita and average incomes
In principle there are three equivalent ways to calculate GDP:
- Production: the value of final outputs produced in an economy less the value of all the inputs used in their production
- Expenditure: the expenditure on final goods and services produced in an economy by households, corporations and governments
- Income: the income earned by individuals and businesses from the production of goods and services in the economy
It theory these three measures should be equal; they constitute an accounting identity. A company’s revenue is the income it generates from selling the goods and services it produces to consumers; yet that same revenue is also the expenditure of consumers on those goods and services. Paul Krugman makes the point that “our income mostly comes from selling things to each other. Your spending is my income, and my spending is your income.”7 This symmetry allows us to use GDP to approximate average incomes by dividing GDP by the total population. In reality, average incomes and GDP per capita will not be equal.8
# Real GDP: adjusting for inflation in intertemporal comparisons
Nominal GDP is a measure of the value of output produced in a country or region over a specified period (usually one year). The value of output is composed of two factors: the volume produced and the price,
where represents the price of output and represents the volume of output. Therefore increases to GDP are either the result of more output, higher prices or a combination of the two. When looking at the performance of a single economy over time it is important that we control for price effects since they can mask changes to the value of output. In these cases we adjust for the price changes and look at real GDP. The real GDP is constructed by ‘deflating’ the nominal GDP by a price index that tracks changes to prices in the economy relative to a chosen base year. This transformation attempts to isolate volume changes by eliminating price effects.
However, it is not just prices that change over time. The very products and services that we produce and buy change. Technological progress has meant that the goods and services available today are invariably superior to those available several hundred years ago, with almost no example to the contrary. The introduction of new goods and services creates serious problems for intertemporal comparisons of wealth that are most relevant today; it is less of a problem for the long pre-modern world when almost all economic production consisted of food, shelter and clothing. To emphasise this point consider the following example: In 1836 the richest man in the world was probably Nathan Rothschild. Rothschild could afford whatever he wanted – every good and service available in the world of 1836. Yet in that same year, the 56 year old died of an infection that is curable today by cheap antibiotics. 10
# Nominal wages, consumer prices, and real wages in the UK since 1750
# Quality changes
The Stiglitz, Sen, Fitoussi report explains “Quality change can be very rapid in areas like information and communication technologies. There are also products whose quality is complex, multi-dimensional and hard to measure, such as medical services, educational services, research activities and financial services. Difficulties also arise in 1. Evidence and references in support of the claims presented in this Summary are presented in a companion technical report. 22 SHORT NARRATIVE ON THE CONTENT OF THE REPORT collecting data in an era when an increasing fraction of sales take place over the internet and at sales as well as discount stores. As a consequence, capturing quality change correctly is a tremendous challenge for statisticians, yet this is vital to measuring real income and real consumption, some of the key determinants of people’s well-being. Under-estimating quality improvements is equivalent to over-estimating the rate of inflation, and therefore to underestimating real income. For instance, in the mid-1990s, a report reviewing the measurement of inflation in the United States (Boskin Commission Report) estimated that insufficient accounting for quality improvements in goods and services had led to an annual overestimation of inflation by 0.6%. This led to a series of changes to the US consumer price index.”11
The two most common price indices used to deflate incomes and nominal GDP are:
- Consumer Price Index (CPI): The basket is used measure the price changes of the goods and services consumed by the typical household. The typical household consumption basket will vary from country-to-country due to different preferences and income levels as well as over time as new technologies emerge and preferences change. CPI indices are produced by national statistical agencies, with one major exception being the European Union’s Harmonised CPI (HCPI). The HCPI uses consistent definitions for all Eurozone members to aid comparability.The CPI used to adjust household incomes tracks changes to the price of a basket of goods that the typical household purchases. It is measuring the price changes of consumption.
- GDP deflator: The GDP deflator also called National Accounts deflator measures the purchasing power relative to the prices of all domestically produced final goods and services in an economy. It is measuring the price changes of domestic production.
# Differences between CPI and GDP deflator
The CPI index measures price changes of consumption whereas the GDP deflator measures price changes of domestic production.
Consequently there are several important differences of the two price indices:12
- Unlike the CPI, the GDP deflator does not adjust for changes of goods that are imported from other countries and the price changes of import prices are not directly taken into account.
- Secondly, the GDP deflator covers capital goods, goods that are not bought by consumers.
- The CPI is also available monthly for most countries, while the GDP deflator is mostly available only quarterly.
# Composition of the bundle of goods and services measured in the CPI
The following chart displays the composition of the German CPI basket of goods. The basket used is chosen to reflect the expenditure of the typical household, and changes to the CPI index over time reflect changes to the prices the typical consumer will face. The typical household consumption basket will vary from country-to-country due to preferences as well as over time as new technologies emerge and preferences change.13
# Composition of the German CPI basket of goods
# International GDP Comparisons: market vs. PPP exchange rates
In order to highlight the difficultly of making international comparisons between countries, consider the average income of somebody living in India compared with the US.
In 2011 the average Indian earned 72,000 rupees, while the average American earned $50,000. Since the average incomes are stated in different currencies, comparing the numerical value is meaningless and does not help us determine who is richer and by how much. Converting the rupee amount into US dollars using market exchange rates gives us an average income of $1,500 in India. This number is over 33 times smaller than the income of the average American!
However, it is obvious that the cost of living in the US is much higher than in India, which implies that the comparison of incomes made at market exchange rates is also not a fair comparison of how rich or poor the people really are in comparison.14 A solution is to convert the amounts using the Purchasing Power Parity (PPP) exchange rate. This conversion takes into account differences in the price levels of both countries. Making the PPP-adjustment reveals that the average income of someone living in India is $4,800 (international dollars), only 10 times smaller than in the US. The precise nature of PPP adjustments is explained in the section below.
This does not mean that comparisons of GDP evaluated at market exchange rates are uninformative. In fact, when comparing financial flows, PPP-adjustments are meaningless and GDP evaluated at the market exchange rate is the most appropriate measure. When comparing development and living standards, the converse is true since we need to eliminate price effects.
# PPP-adjusted GDP: spatial comparisons
GDP comparisons made using market exchange rates fail to reflect differences in the purchasing power of different currencies. In general, prices are higher in developed economies,15 and so exchange rate adjusted GDP measures will underestimate the size of low income economies.
In a simplified world, the prices of traded goods are determined by global demand and supply forces, while the prices of non-traded goods are determined by local demand and supply forces. Since wages and salaries are lower in developing countries, the prices of non-traded goods also tend to be lower. This feature is missed by exchange rate adjusted GDP calculations as no distinction is made between traded and non-traded goods. In a world where all goods are traded, exchange rate adjusted GDP would be a more informative way to make international comparisons. Nevertheless, another complicating factor is that exchange rates are highly volatile and determined by currency speculation, interest rates and international capital flows.
Purchasing Power Parity (PPP) adjustments to GDP are an attempt to isolate differences in the volume of output of two economies. That is, they eliminate disparities in the price levels of different economies. A PPP exchange rate can be thought of as the cost ratio of a comparable (but not identical) basket of goods in two countries. Hence, the methodology is analogous to that used in computing CPI inflation to calculate real GDP, except that here the comparisons are made between countries rather than over time. More precisely, PPP-adjusted GDP is a spatial measure and not a temporal one like real GDP: a base country is used as opposed to a base year. The US dollar is the most common unit of currency used to make international comparisons and, for clarity, PPP-adjusted quantities are quoted in international or Geary-Khamis dollars.16
We can decompose the GDP ratio of two economies into
where is the price level in country , is the currency level in country , and is the volume of output (real output) in country . When making international comparisons we are interested in the ratio of output volume. It is possible to remove the currency differences by using the exchange rate to convert the GDP into a common currency, but this would leave price level differences. The PPP exchange rate adjusts for both the currency and the price level ratio.
# The creation of PPPs by the International Comparisons Programme (ICP)
The PPPs that are used in international comparisons today are created by the International Comparisons Programme (ICP) conducted by the World Bank. The latest round of the ICP was completed in 2014 and has estimated PPPs for 2011. The study covers 199 countries and is the most extensive study of PPPs ever conducted.
Since the research is highly intensive and requires cooperation with many different countries and statistical agencies, the existing PPP data is sparse (8 benchmark years since the first ICP study in 1970). What is more, the methodology used and the group of participating countries has differed between each round of the ICP. There are six years separating the most recent ICP estimates (2011 and 2005). Her is a detailed explanation of the methodology and findings of the 2011 ICP.
# Computing PPP-adjusted GDP for years for which no ICP data is available
Computing PPP-adjusted GDP for years where ICP data is available is straightforward, however, in years where there is no data, there is no consensus regarding the best way to produce estimates.
Since PPP is a spatial measure, each ICP estimate is indexed to a benchmark year. Producing PPP-adjusted GDP estimates for non-benchmark years requires either extrapolation of PPP-estimates from a single round of ICP data, interpolation between different rounds or a combination of the two. Extrapolation takes the PPP-adjusted GDP in a single year and assumes that it evolves according to real GDP growth rates or the inflation ratio of the country of interest with the US. Interpolation attempts to “fill-in the gaps” using observed ICP rounds in conjunction with inflation or growth data according to some statistical model. The two figures presented below are designed to aid understanding of these two methods.
Understanding the methodology and purpose of each dataset is important when conducting data analysis. The four most important data sets for PPP-adjusted GDP data have the following methodology:
- World Bank data: Extrapolates from the most recent ICP round using national accounts deflators. The data and a brief description of the methodology can be found at http://data.worldbank.org/indicator/PA.NUS.PRVT.PP.
- Since the data is extrapolated from the 2011 ICP, data is only presented for the years 1990-2014 as any further extrapolation is likely to give unreliable estimates. This dataset is most useful when considering the very recent past as the 2011 ICP round uses the most sophisticated methodology to date. It also has the advantage of being extrapolated forward in time to 2014.
- Notes from the documentation: “For most economies PPP figures are extrapolated from the 2011 International Comparison Program (ICP) benchmark estimates or imputed using a statistical model based on the 2011 ICP.”
- Penn World Tables: Interpolates and extrapolates using national accounts deflators. The methodology used can be found in the documentation available at http://www.rug.nl/research/ggdc/data/pwt/pwt-8.1.
- This dataset is most commonly used for statistical analysis by economists and covers the years 1950-2011. It is arguably the most reliable, long-run data available on PPP-adjusted GDP.
- Gapminder: Interpolates and extrapolates using real growth rates. The methodology used can be found in the documentation.
- Gapminder aims to give the widest coverage possible at the expense of robust estimates. Much of the historic data has been estimated from trends and other For this reason, we present the data here for the purposes of graphical presentation and not as fact.
- Notes from the documentation: “The main purpose of the data is to produce graphical presentations that display the magnitude of income disparities in the world over time… Hence we discourage the use of this data set for statistical analysis… The observations for the period before 1950 are, in the majority of cases, based on rough estimates within a range of likely values. In many cases we have no information, what-so-ever, on the relative ranking of countries.”
- Maddison Project: Data is drawn from various sources. A complete list of sources can be found in the documentation available at http://www.ggdc.net/maddison/maddison-project/home.htm.
- This dataset has many different sources and is curated by the project team. Although the long-run data is much less reliable than more recent estimates, the more rigorous nature of the estimates allows for better spatial comparisons.
- Notes from the documentation: “The Maddison Project database places the emphasis more on the international comparability of the estimates than on their consistency over time.”
# Discrepancy between incomes reported in household surveys and GDP per capita
This discrepancy reflects several factors:
– GDP includes items such as depreciation, retained earnings of corporations, and government revenues that are not distributed back by the government or corporations to households as cash transfers.
– Particularly incomes at the very top of the income distribution are not fully accounted and this contributes to the gap between the National Account figure (GDP) and household survey figures.
– Particularly in developing countries – where this discrepancy tends to be larger – a large share of government revenue may firstly end up in sovereign wealth funds or secondly represents profits of foreign multinationals that are repatriated but taken into account in the GDP calculation.
See the Appendix of the original publication for a longer explanation.[/ref]
# Data Sources
# Long-run datasets
# Big datasets
# Angus Maddison Historical Statistics
- Data: GDP per capita and total GDP (+ population data)
- Geographical coverage: Global – by countries & world regions
- Time span: 1-2008 CE (5 observations before 1820 and then annual if possible)
- Available at: Online at the Groningen Growth & Development Centre here
- Founded by the economic historian Angus Maddison and later updated. An updated version – available through the CLIO Infra project – is described below. Another correction of the Maddison data is the ‘Barro-Ursua Macroeconomic Data’, which is available at Robert Barro’s website here.
# Maddison Project – available through the CLIO Infra Project
- Data: GDP per capita
- Geographical coverage: Global coverage – by country
- Time span: Between 1500 and 2010
- Available at: Online here
- This database builds on the work of Angus Maddison. The database and the accompanying paper17 are a product of The Maddison Project and are partly a revision of the earlier work. The authors are Jutta Bolt and Jan Luiten van Zanden. The many sources are very well documented at the website of the CLIO project.
# The Global Price and Income History Group
- Data: Nominal GDP
- Geographical coverage: Many world regions (including Africa, Asia and Oceania)
- Time span: Varies by world region – long for some (e.g. Italy since 1310) but mostly since the 19th century
- Available at: Online here
- Well documented, readily available in Excel spreadsheets. Includes many other datasets
# Small Datasets
Wikipedia’s ‘list of regions by past GDP (PPP) per capita’ includes Maddison’s estimates for countries and world regions between 1CE and 2003CE. It also includes Bairoch’s estimates for Europe between 1830–1938. And some estimates for the prosperity of the Roman and Byzantine empires.
# Stephen Broadberry’s Data
- Data: GDP per capita
- Geographical coverage: Britain, Italy, Spain and Holland 1270-1870 and Europe for 1870-2000
- Time span: 1270-1870 and 1870-2000
- Available at: The early data is published in research papers by Broadberry and others. An overview is given in Broadberry (2013)18 The data for several European countries for the time period 1870–2000 is available online at Broadberry’s website here.
- The quality of this data is high. The data was recently published in detailed research papers by various authors.
# National Accounts data in the ‘International Historical Statistics’
- Data: GDP per capita and some other National Accounts data
- Geographical coverage: Global – by country
- Time span: Varies by country. Often since the early 19th century (US since 1789)
- Available at: The statistics are published in three volumes covering more than 5000 pages.19 At some universities you can access the online version of the books where data tables can be downloaded as ePDFs and Excel files. The online access is here.
- These statistics – originally published under the editorial leadership of Brian Mitchell (since 1983) – are a collection of data sets taken from many primary sources, including both official national and international abstracts dating back to 1750.
# Jerry Dwyer’s data set
- Data: National account data & other relevant data (e.g. average age and experience of the workforce)
- Geographical coverage: Global by country
- Time span: Mostly since 1880 – observations at every even decade
- Available at: Online here. Downloadable as an excel-file
# The Historical National Accounts database of the Groningen Growth and Development Centre
- Data: National Account data with many interesting measures
- Geographical coverage: (Mostly but not only) Early industrialized countries
- Time span: Since the early 19th century
- Available at: It is online here.
# The History Database of the Global Environment (HYDE)
- Data: GDP (per capita) and private consumption as share of GDP
- Geographical coverage:
- Time span: Since 1890
- Available at: GDP data is online here. Private consumption is here.
- Early estimates are based on Maddison’s data.
# Penn World Table
- Data: GDP and many measures related to relative levels of income, output, inputs and productivity.
- Geographical coverage: Global – by country
- Time span: Since 1950
- Available at: It is online here.
- The documentation can be found in Summers and Heston (1991)20 and the follow-up papers.
- For a discussion of the validity of these measures see Daniel A. Nuxoll (1994), Robert C. Feenstra, Alan Heston, Marcel P. Timmer, And Haiyan Deng (2009) and Simon Johnson, William Larson, Chris Papageorgiou, Arvind Subramanian (2013)21
# World Development Indicators (WDI) published by the World Bank
- Data: GDP and GNI (gross national income) constructed and corrected for price differences (across time and countries) in different ways
- Geographical coverage: Global – by country and world region
- Time span: The earliest data is available for 1960.
- Available at: The different measures are published as part of the World Development Indicators. The data on current GDP in US-$ is here.
# Total Economy Database (TED)
- Data: GDP, population, employment, hours, labor quality, capital services, labor productivity, and total factor productivity
- Geographical coverage: Global – by country
- Time span: Since 1950
- Available at: It is online here.
# Other GDP-related data
# Subnational data
Nicola Gennaioli, Rafael La Porta, Florencio Lopez-de-Silanes, and Andrei Shleifer present subnational data of income in their publication22 The data covers 1,569 subnational regions from 110 countries covering 74% of the world’s surface and 97% of its GDP.
# Spatial Data
Geographically based Economic data (G-Econ) by William Nordhaus and Xi Chen. The dataset covers “gross cell product” for all regions for 1990, 1995, 2000, and 2005 and includes 27,500 terrestrial observations. The basic metric is the regional equivalent of gross domestic product. Gross cell product (GCP) is measured at a 1-degree longitude by 1-degree latitude resolution at a global scale.