The classical tradition running from Locke and Smith through North's institutional economics, which treat enforceable claims to property as the enabling condition for capital investment.
Methodology: How we measure
Designing the Index
We based the construction of the Global Capitalism Index (GCI) on one principle and a systematic survey of the empirical and theoretical literature. We contend that capitalism is not simply pure "economic freedom" or the absence of government. Instead, the relationship between government and capitalism resembles a convex parabolic curve (see Figures 1 and 2 below) in which economies can become more capitalist as they move away from minimal government activity, up to a point.
To be clear, state involvement can and often does impede capitalist systems past a moderate point of what Alexander Hamilton described as "limited but energetic government." Of course, individual metrics can have a linear, downward-sloping relationship between government and capitalism—state-owned enterprises are nearly always "anti-capitalist." Nevertheless, we maintain that any constraints on the ability to deploy capital in open markets are anti-capitalist, regardless of whether they come from government or the private sector.
To be clear, state involvement can and often does impede capitalist systems past a moderate point of what Alexander Hamilton described as "limited but energetic government." Of course, individual metrics can have a linear, downward-sloping relationship between government and capitalism—state-owned enterprises are nearly always "anti-capitalist." Nevertheless, we maintain that any constraints on the ability to deploy capital in open markets are anti-capitalist, regardless of whether they come from government or the private sector.
Figure 1: "Economic Freedom" Index Model
Figure 2: Global Capitalism Index Model
The GCI's eight subindex structure emerged from a comprehensive review of the scholarly literature on capitalism and its related fields. Beginning with Adam Smith and other classical political economists, our study spanned the nineteenth-century work of Mill, Marx, George, and Sumner. We included the twentieth-century scholarship ranging from Schumpeter, Hayek, and Polanyi and concluded with the recent institutional economics of North, Acemoglu, Rodrik, Hall and Soskice, and many more.
The eight subindices represent empirical constructions of vital elements of capitalist systems identified in the literature.
Each subindex has its roots in a distinct theoretical tradition:
The eight subindices represent empirical constructions of vital elements of capitalist systems identified in the literature.
Each subindex has its roots in a distinct theoretical tradition:
Each of these eight concepts can be broken down into more granular areas of study, but this level of decomposition maps onto the theoretical traditions that have shaped how scholars think about capitalism for the last two centuries. They also can be measured and proxied through publicly available datasets. The GCI Whitepaper presents the full theoretical case for each subindex, along with alternative models that we considered and rejected.
How the Global Capitalism Index's Approach is Different
For decades, public, private, and academic actors have failed to reach consensus on exactly what capitalism is and how it operates. Furthermore, the existing indices used by scholars to study capitalism explicitly measure "economic freedom," not capitalism itself. We believe these indices are methodologically and definitionally imprecise, making a systematic study of capitalism virtually impossible.
The status quo assumption that "less government" is the definitive feature of a capitalist economy lacks historical grounding and theoretical rigor. Yet the methodological problems compound the definitional ones. Widely used indices rely on simple averages across datasets and/or arbitrary weighting to score countries. The result is that the current products used to measure capitalism do not accurately assess capitalist systems in whole or in part. The GCI provides the scope necessary to investigate each element of capitalism—and its comprehensive whole—in depth. We also hope that the GCI will provide a lingua franca through which we can debate capitalism's merits and effect on society. Please also see our What is Capitalism? page.
The status quo assumption that "less government" is the definitive feature of a capitalist economy lacks historical grounding and theoretical rigor. Yet the methodological problems compound the definitional ones. Widely used indices rely on simple averages across datasets and/or arbitrary weighting to score countries. The result is that the current products used to measure capitalism do not accurately assess capitalist systems in whole or in part. The GCI provides the scope necessary to investigate each element of capitalism—and its comprehensive whole—in depth. We also hope that the GCI will provide a lingua franca through which we can debate capitalism's merits and effect on society. Please also see our What is Capitalism? page.
Collecting and Normalizing the Data
Every GCI score is the final product of an extensive data selection, collection, and normalization process:
Fitting the Pieces Together
Every weight in the GCI emanates from the data rather than an arbitrary value assigned by the research team. We use multi-tiered adaptive principal component analysis (PCA) to derive those weights. The PCA process identifies how a group of indicators (datasets) move together and weights each one according to how strongly it aligns with that shared pattern. Indicators that track the underlying shared pattern receive a higher weight, while indicators that move with more autonomy receive a lower weight. In an arbitrary weighting scheme, a simple average treats redundant datasets as independent evidence, so three indicators measuring nearly the same thing will outvote one measuring something distinct. PCA recognizes that overlap and assigns lower weights to those redundant datasets accordingly.
We apply the same logic at both levels of our measurement design. PCA builds each subindex from its contributory datasets, with source reliability grades shaping how much influence each dataset carries. Then, a second PCA combines the eight subindices into the composite. Some subindices consist of thinner data than others, and a less precise measure will appear less connected to the rest than it truly is. Thus, we use each subindex's confidence interval to account for deficient precision before the PCA process sets the weights. A subindex earns its weight from the concept it captures rather than being penalized for gaps in the data behind it.
We apply the same logic at both levels of our measurement design. PCA builds each subindex from its contributory datasets, with source reliability grades shaping how much influence each dataset carries. Then, a second PCA combines the eight subindices into the composite. Some subindices consist of thinner data than others, and a less precise measure will appear less connected to the rest than it truly is. Thus, we use each subindex's confidence interval to account for deficient precision before the PCA process sets the weights. A subindex earns its weight from the concept it captures rather than being penalized for gaps in the data behind it.
Read More
For the full technical methodology, subindex construction, data sources, and validity analysis, please see the GCI Whitepaper. For country-level scores, subindex breakdowns, and interactive comparisons, explore the data on our GitHub.
FAQs
Capitalism is a system that privileges the ability to deploy capital in open markets as freely as possible, while minimizing constraints from both public and private sources.
The "as freely as possible" qualifier is important. A system can be capitalist even if "the ability to deploy capital in open markets" is not absolute. The question is about degrees—do prospective obstacles prevent (not just make harder in some circumstances) the ability to "deploy capital in open markets." Of course, we measure the effect of those impediments in the aggregate GCI score, allowing users to see how well a capitalist system functions at a given time. For a fuller explanation of our definition, see What is Capitalism?
The "as freely as possible" qualifier is important. A system can be capitalist even if "the ability to deploy capital in open markets" is not absolute. The question is about degrees—do prospective obstacles prevent (not just make harder in some circumstances) the ability to "deploy capital in open markets." Of course, we measure the effect of those impediments in the aggregate GCI score, allowing users to see how well a capitalist system functions at a given time. For a fuller explanation of our definition, see What is Capitalism?
By "capital" we mean the three widely recognized types of assets or resources used to make future work or production more efficient and/or profitable. These are:
1. Financial Capital: Cash, equities, bonds, or other financial instruments used to fund business activities or investments.
2. Physical Capital: Tangible tools, machinery, buildings, and equipment used to produce goods or services.
3. Human Capital: Knowledge, skills, education, experience, and abilities people use to create economic value.
We contend that all three forms of capital hold equal ex ante value. Traditional definitions of capitalism tend to favor financial or physical capital, but empirical studies suggest that human capital is equally important to the process of value creation. Thus, we hold the ability for a person to "deploy their human capital in open markets as freely as possible" to be just as important for a well-functioning capitalist system as any other.
1. Financial Capital: Cash, equities, bonds, or other financial instruments used to fund business activities or investments.
2. Physical Capital: Tangible tools, machinery, buildings, and equipment used to produce goods or services.
3. Human Capital: Knowledge, skills, education, experience, and abilities people use to create economic value.
We contend that all three forms of capital hold equal ex ante value. Traditional definitions of capitalism tend to favor financial or physical capital, but empirical studies suggest that human capital is equally important to the process of value creation. Thus, we hold the ability for a person to "deploy their human capital in open markets as freely as possible" to be just as important for a well-functioning capitalist system as any other.
Most conventional frameworks posit government and capitalism as having a strong inverse relationship. This framework results in the assumption that state activity, such as higher taxes or more generous social safety nets, automatically diminish a country's capitalist bona fides.
Yet a true capitalism index must account for nuance: are we talking about income taxes? Corporate taxes? VATs? Tariffs, or overall tax burden? When considering the presence of safety nets, are we examining the type of universal health coverage, or measuring its impact on entrepreneurial activity? The GCI measures capitalism as an institutional system separate from fiscal and political outcomes. Put another way, the GCI examines how a country makes money, not what it spends that money on.
To be clear, there are many ways, both in type and degree, that state activity impedes or hampers capitalism. We simply refuse to take any activity as an ex ante negative—what the state does, and the degree to which it does it, matters immensely. We try to capture those details as opposed to drawing stark, unexamined lines.
Yet a true capitalism index must account for nuance: are we talking about income taxes? Corporate taxes? VATs? Tariffs, or overall tax burden? When considering the presence of safety nets, are we examining the type of universal health coverage, or measuring its impact on entrepreneurial activity? The GCI measures capitalism as an institutional system separate from fiscal and political outcomes. Put another way, the GCI examines how a country makes money, not what it spends that money on.
To be clear, there are many ways, both in type and degree, that state activity impedes or hampers capitalism. We simply refuse to take any activity as an ex ante negative—what the state does, and the degree to which it does it, matters immensely. We try to capture those details as opposed to drawing stark, unexamined lines.
The GCI is comprised of 213 datasets drawn from public, private sector, and academic sources including the World Bank, the International Monetary Fund, the Varieties of Democracy Project (V-Dem), Bloomberg, and others. Several appendices in the GCI Whitepaper document every dataset used in the GCI, including a description of the dataset, its original scale, our normalization method, its reliability grade, and a hyperlink to its original source.
Principal component analysis (PCA) creates measurement weights from the shared movements across a group of datasets. These weights then determine how much each dataset contributes to a subindex score. No single source of information can fully observe a complex and abstract concept like "property rights" or "market competition." Thus, experts must weave together individual datasets that, when combined in the right way, can provide a sophisticated measurement of the underlying concept. Countries that score high on one dataset tend to score high on the others, and when a country's value on one rises over time, its values on the others tend to rise with it. When several datasets move together, the most likely explanation is that they are all responding to the same institutional quality, and PCA weights each dataset by how closely it follows that shared movement.
We apply PCA adaptively by recalculating the weights whenever the available datasets change. Over the course of a two-decade study period, new sources launch, old datasets stop publishing results, and existing sources revise their methods. Thus, rather than fixing our weights in 2009 and never adjusting them, we combine the years into coverage periods or groups of consecutive years in which the data pool remains unchanged. For each of those periods, we run the PCA calculation to produce one set of weights specific to those years. When the data pool changes, we calculate a new set of weights that accurately represent the underlying data.
Since separate periods produce separate weights, their scores are not directly comparable until we link them. Linking means multiplying every score in the later period by one number and adding another to it, chosen so that countries scored in both the last year of the earlier period and the first year of the later one have the same average score in both years. Without this step, a country's score would shift sharply in a year when nothing about that country changed simply because a new dataset entered the pool. Since the same adjustment applies to every country in the period, it removes that shift without altering how countries rank against each other or the distance between their scores. Thus, the model still measures genuine institutional change and the full series reads on a single continuous 0–100 scale.
We apply PCA adaptively by recalculating the weights whenever the available datasets change. Over the course of a two-decade study period, new sources launch, old datasets stop publishing results, and existing sources revise their methods. Thus, rather than fixing our weights in 2009 and never adjusting them, we combine the years into coverage periods or groups of consecutive years in which the data pool remains unchanged. For each of those periods, we run the PCA calculation to produce one set of weights specific to those years. When the data pool changes, we calculate a new set of weights that accurately represent the underlying data.
Since separate periods produce separate weights, their scores are not directly comparable until we link them. Linking means multiplying every score in the later period by one number and adding another to it, chosen so that countries scored in both the last year of the earlier period and the first year of the later one have the same average score in both years. Without this step, a country's score would shift sharply in a year when nothing about that country changed simply because a new dataset entered the pool. Since the same adjustment applies to every country in the period, it removes that shift without altering how countries rank against each other or the distance between their scores. Thus, the model still measures genuine institutional change and the full series reads on a single continuous 0–100 scale.
We combine the eight subindices into a composite GCI score through a second PCA, operating on the same principles as the first. Rather than producing a simple average across all subindices, we measure the degree to which countries scoring higher on one subindex tend to score higher on the others and thereby summarize the shared movement as a single underlying pattern. Each subindex is weighted by how closely it tracks that shared movement. That movement forms a single scale running from countries that score low on all eight subindices to countries that score high across all eight. A country's composite score is where it falls on that scale. A single shared pattern accounts for about three-quarters of what distinguishes one country from another across the eight subindices. The remaining quarter represents the autonomous structure of a series of data, such as a country that scores well overall but lags in one subindex. Our publication of all eight subindices alongside the composite GCI score allows users to study these relationships between all elements of the index.
Where a dataset behind a subindex is missing a value for a country in a given year, we fill that gap with estimated values (see next FAQ on how we deal with confidence intervals and missing data). When two countries' scores on a subindex differ, part of that difference reflects the true difference between them and part reflects the values we have estimated. The confidence interval measures how much of a score is comprised of estimated rather than true, observed values. A wide confidence interval means that much of the gap between two countries could come from estimated values rather than from any real difference between them. When the PCA process calculates weights, it ignores the differences between countries that could have arisen from estimation error. Put another way, our weights only reflect the true observed differences between countries, or the reported differences between countries minus the confidence interval. For example, since fewer countries report the datasets behind the Free Flow of Goods and Capital subindex we must estimate more of its values, and thus its confidence interval is wider. If we did not adjust for the wider confidence interval, Free Flow of Goods and Capital would receive a smaller weight simply because more of its values had to be estimated, not based on its importance to a country's capitalist system.
Where a dataset behind a subindex is missing a value for a country in a given year, we fill that gap with estimated values (see next FAQ on how we deal with confidence intervals and missing data). When two countries' scores on a subindex differ, part of that difference reflects the true difference between them and part reflects the values we have estimated. The confidence interval measures how much of a score is comprised of estimated rather than true, observed values. A wide confidence interval means that much of the gap between two countries could come from estimated values rather than from any real difference between them. When the PCA process calculates weights, it ignores the differences between countries that could have arisen from estimation error. Put another way, our weights only reflect the true observed differences between countries, or the reported differences between countries minus the confidence interval. For example, since fewer countries report the datasets behind the Free Flow of Goods and Capital subindex we must estimate more of its values, and thus its confidence interval is wider. If we did not adjust for the wider confidence interval, Free Flow of Goods and Capital would receive a smaller weight simply because more of its values had to be estimated, not based on its importance to a country's capitalist system.
We publish every GCI score with a 95% confidence interval, meaning a range within which the country's true score very likely falls. Since some countries do not report every value for every dataset in every year, we must fill in those missing numbers by using that country's own values in adjacent years or by borrowing a value observed for the countries whose other measurements most resemble it. In so doing, we make an informed but imperfect estimate to fill in for the missing value. Of course, the nature of these estimations means that our score for a given country may differ from the true value if we had perfect data. The confidence interval provides a range within which we feel very confident the true score lies.
To calculate the confidence interval, we run 500 simulations of each published score. Each simulation takes a different sample of country-years and fills each missing entry with a different plausible value. It then records how far the score rises or falls. The middle 95% of those 500 results is the range we publish. A country whose data is complete still has a range, because the countries it is compared against change with each simulation.
When two countries' ranges do not overlap, one genuinely ranks above the other. When the ranges do overlap, the data cannot tell us which country is truly higher because the difference between them is small enough that it could come from data fallibility or methodological change.
To calculate the confidence interval, we run 500 simulations of each published score. Each simulation takes a different sample of country-years and fills each missing entry with a different plausible value. It then records how far the score rises or falls. The middle 95% of those 500 results is the range we publish. A country whose data is complete still has a range, because the countries it is compared against change with each simulation.
When two countries' ranges do not overlap, one genuinely ranks above the other. When the ranges do overlap, the data cannot tell us which country is truly higher because the difference between them is small enough that it could come from data fallibility or methodological change.
The GCI scores range from 0–100, with scores closer to 100 indicating stronger alignment with the concept measured in each subindex. In practice, the relative difference between country scores and change over time are the most useful comparisons. These allow us to see which countries meaningfully improve or decline, and what other countries present similar results. Typically, a country scoring a 90 is in the top handful of countries for that concept, and a country scoring a 50 is towards the middle to low end of the global distribution.
At some level, every measurement depends on the choices or judgements made by its creators. The GCI is no different. However, our theoretical models are the product of a comprehensive evaluation of the academic and scholarly literature across time and disciplines. Furthermore, our measurement methodology deploys the most advanced analytical techniques to avoid, so much as is possible, ideological bias and technical error.
Most importantly, we have built theoretical and methodological transparency into every element of the GCI. We meticulously document every step of our data sourcing, transformation and harmonization, index composition, testing and validation, and analytical modeling in our GCI Whitepaper and other documentation available here. Since all data are public, we strongly encourage any user to inspect, replicate, or contest our results.
Most importantly, we have built theoretical and methodological transparency into every element of the GCI. We meticulously document every step of our data sourcing, transformation and harmonization, index composition, testing and validation, and analytical modeling in our GCI Whitepaper and other documentation available here. Since all data are public, we strongly encourage any user to inspect, replicate, or contest our results.
The index begins in 2009 because that is the earliest year with adequate cross-country data coverage across all 213 datasets. The GCI will be updated annually.
The GCI and its subindices cover the period 2009–2025, with each edition designated by the year following the calendar year to which the underlying data pertain. Accordingly, the 2025 edition of the GCI reflects institutional conditions in calendar year 2024, rather than in 2025 itself. This convention has important implications for empirical analysis. In particular, when assessing whether the GCI responds to a specific world event, one should subtract one year from the edition label to identify the relevant year of observation.
The GCI provides a basis for measuring and debating capitalist economic systems. We anchor these models to a systematic theoretical process and absolute empirical and technical rigor.
For researchers, the GCI provides a comprehensive and rigorous tool to evaluate the effect of capitalist institutions on subjective wellbeing, economic development, and institutional design. The GCI's composition from 8 independent subindices and overt separation from outcome variables allows scholars to study capitalism as a whole, or any of its 8 individual components, with minimal structural bias.
For policymakers, the availability of 8 "free-standing" subindices allows for the identification of previously unobservable strengths and weaknesses in a nation's institutional structure. Since specific elements of the GCI (domains and subindices) have unique, statistically significant relationships with a wide range of output variables, we hope that policy makers can use the GCI to better target reforms and other policy decisions.
For public understanding, the GCI reveals that many countries considered "less capitalist" often have stronger capitalist institutions than nations commonly thought of as "more capitalist." We hope to empower users to consider the nuance of capitalist systems and better understand those systems' strengths and weaknesses.
For researchers, the GCI provides a comprehensive and rigorous tool to evaluate the effect of capitalist institutions on subjective wellbeing, economic development, and institutional design. The GCI's composition from 8 independent subindices and overt separation from outcome variables allows scholars to study capitalism as a whole, or any of its 8 individual components, with minimal structural bias.
For policymakers, the availability of 8 "free-standing" subindices allows for the identification of previously unobservable strengths and weaknesses in a nation's institutional structure. Since specific elements of the GCI (domains and subindices) have unique, statistically significant relationships with a wide range of output variables, we hope that policy makers can use the GCI to better target reforms and other policy decisions.
For public understanding, the GCI reveals that many countries considered "less capitalist" often have stronger capitalist institutions than nations commonly thought of as "more capitalist." We hope to empower users to consider the nuance of capitalist systems and better understand those systems' strengths and weaknesses.
Data Sources
Every dataset behind the Index, grouped by subindex and domain. Pick a subindex to see its sources.
| Domain | Dataset | Source |
|---|---|---|
| Intellectual Property Protection | International Property Rights Alliance Patent Index | www.internationalpropertyrightsindex.org |
| WEF Executive Opinion Survey: IP Protection | initiatives.weforum.org | |
| Chamber of Commerce International IP Index | www.theglobalipcenter.com | |
| Lack of Corruption | Transparency International CPI | www.transparency.org |
| WGI Control of Corruption | www.worldbank.org | |
| V-Dem Judicial Corrupt Decisions | www.v-dem.net | |
| V-Dem Judicial Purges | www.v-dem.net | |
| V-Dem Government Attacks on Judiciary | www.v-dem.net | |
| V-Dem Court Packing | www.v-dem.net | |
| V-Dem High Court Independence | www.v-dem.net | |
| V-Dem Lower Court Independence | www.v-dem.net | |
| State Stability | Fragile States Index | fragilestatesindex.org |
| WGI Political Stability | www.worldbank.org | |
| WGI Government Effectiveness | www.worldbank.org | |
| Political Constraint Index | mgmt.wharton.upenn.edu | |
| Political Expression | Reporters Without Borders Press Freedom Index | rsf.org |
| WGI Voice and Accountability | www.worldbank.org | |
| V-Dem Freedom of Expression Index | www.v-dem.net | |
| Rule of Law | WJP Rule of Law Index | worldjusticeproject.org |
| WGI Rule of Law | www.worldbank.org | |
| V-Dem Judicial Accountability | www.v-dem.net | |
| V-Dem High Court Compliance | www.v-dem.net | |
| V-Dem Judicial Review of Laws | www.v-dem.net | |
| V-Dem Transparent and Predictable Law Enforcement | www.v-dem.net | |
| V-Dem Access to Justice | www.v-dem.net | |
| Contract Enforcement | World Bank Enforcing Contracts | www.worldbank.org |
| World Bank Resolving Insolvency | www.worldbank.org | |
| CPIA Rule-Based Governance | databank.worldbank.org | |
| Private Ownership | V-Dem State Ownership of the Economy | www.v-dem.net |
| Government Final Consumption Expenditure (% GDP) | data.worldbank.org | |
| Gross Fixed Capital Formation | data.worldbank.org |