Rationale and origins
The Global Innovation Index (GII) was launched in 2007 by Prof. Soumitra Dutta (then at INSEAD) with the aim of identifying and determining metrics and methods that could capture a picture of innovation in society that is as complete as possible.
There were several motivations for setting this goal. First, innovation is important for driving economic progress and competitiveness – for both developed and developing economies. Many governments are putting innovation at the center of their growth strategies. Second, the definition of innovation has broadened – it is no longer restricted to research and development (R&D) laboratories and published scientific papers. The concept of innovation has become more general and horizontal in nature, and now includes social, business model and technical aspects. Last, but not least, recognizing and celebrating innovation in emerging markets is critical for inspiring people – especially the next generation of entrepreneurs and innovators.
Now in its 17th edition, the GII helps to create an environment in which these innovation factors are subject to continual evaluation. It provides a key tool for decision-makers and a rich database of detailed metrics, offering a convenient source of information for refining innovation policies.
Measuring innovation outputs and their impact remains a challenging task, hence great emphasis is placed on measuring the climate and infrastructure for innovation and assessing related outcomes.
Although the final results are presented as a ranking, the primary aim of the GII is to improve the “journey” to more accurate methods of measurement, understanding innovation and identifying targeted policies, good practices and other levers that foster innovation. The rich data metrics, at index, sub-index or indicator level, can be used to monitor performance over time and to benchmark developments against economies within the same region or income group classification.
Defining innovation in the GII
The GII adopts a broad definition of innovation, originally elaborated in the Oslo Manual developed by the Statistical Office of the European Communities and the Organisation for Economic Co-operation and Development (OECD). In its fourth edition, in 2018, the Oslo Manual introduced a more general definition of innovation: “An innovation is a new or improved product or process (or combination thereof) that differs significantly from the unit’s previous products or processes and that has been made available to potential users (product) or brought into use by the unit (process).” (
This update of the Oslo Manual also introduced a series of definitions associated with innovation in business activities and for different types of innovation firms. In this context, innovation translates as improvements made to outcomes in the form of either new goods or new services, or any combination of these. While the GII focuses on a more general definition of innovation, it is important to highlight how these specific definitions capture the evolution of the way in which innovation has been perceived and understood over the past two decades.
Economists and policymakers previously focused on R&D-based technological product innovation, largely produced in-house and mainly in manufacturing industries. Innovation of this nature was executed by a highly educated labor force in R&D-intensive companies. The process leading to such innovation was conceptualized as closed, internal and localized. Technological breakthroughs were necessarily “radical” and took place at the “global knowledge frontier.” This characterization implied the existence of leading and lagging economies, with low- or middle-income economies only able to play “catch-up.”
Today, innovation capability is increasingly seen as the ability to exploit new technological combinations; it embraces the concept of incremental innovation and “innovation without research.” Non-R&D innovative expenditure is an important component of reaping the rewards of technological innovation. Interest in understanding how innovation evolves in low- and middle-income economies is increasing, along with an awareness that incremental forms of innovation can impact development, and that innovation occurs in the informal economy of developing countries, too (
Furthermore, the process of innovation itself has changed significantly. Investment in innovation-related activity and intangible assets has intensified consistently at the firm, economy and global levels, adding both new innovation actors from outside high-income economies and non-profit actors. The structure of knowledge production activity is more complex, collaborative and geographically dispersed than ever.
Since its inception, the GII has also made a special effort to cover creativity and creative outputs, taking a fresh view of the previously siloed approach to innovation versus creativity. In the opinion of the GII Editors, innovation and creativity are simply two faces of the same coin.
A key challenge is to find metrics that capture innovation as it actually happens in the world today. Direct official measures that quantify innovation outputs remain extremely scarce. For example, there are no official statistics on the amount of innovative activity – defined as the number of new products, processes or other innovations – for any given innovation actor, let alone for any given country. Most measurements also struggle to appropriately capture the innovation outputs of a wider spectrum of innovation actors, such as users or the public and services sectors, or more informal means, which are often the drivers of innovation in developing countries.
The GII aims to improve the measurement of innovation in order to provide a more complete picture of innovation ecosystems across the globe. It explores new metrics regularly to reflect the changing nature of innovation and the increasingly sprawling field of new (big data) innovation indicators.
Interest in applying the GII framework and indicators to develop complementary and mutually reinforcing sub-national innovation indices is also growing among WIPO member states.
The GII conceptual framework
The overall GII ranking is based on two sub-indices that are both equally important in presenting a complete picture of innovation: the Innovation Input Sub-Index and the Innovation Output Sub-Index. Hence, three indices are calculated:
Innovation Input Sub-Index: Five input pillars capture elements of the economy that enable and facilitate innovative activities. The idea is that the innovation inputs of today – and corresponding efforts to develop the science, innovation and human capital base, and the associated innovation environment – prepare the ground for the innovation outputs of tomorrow.
Innovation Output Sub-Index: Innovation outputs are the result of innovative activities within the economy. Although the Output Sub-Index includes only two pillars, it carries the same weight as the Input Sub-Index in calculating the overall GII scores. In other words, innovation output pillars and indicators have a disproportionally greater weight compared to innovation inputs.
The overall GII score is the average of the Input and Output Sub-Indices, from which the GII economy rankings are produced.
Each of the five input and two output pillars is divided into three sub-pillars, each of which is composed of individual indicators – a total of 78 this year (see the Economy profiles section for the Framework of the Global Innovation Index 2024). Each sub-pillar is calculated by taking the weighted average of its individual indicators’ scores, which are normalized to again produce scores between 0 and 100. Pillar scores are calculated using the weighted average of each pillar’s sub-pillar scores.
When WIPO became the sole editor of the GII, the development of a robust and modern data infrastructure was part of the larger plan for GII development, in view of increasing the data quality and data quality control, and the robustness and replicability of the GII model (Appendix Box 1).
To facilitate and permit a comprehensive workflow of the GII model, from data storage to the GII calculations, a robust data infrastructure was developed in 2021 and improved progressively since. The data infrastructure comprises three parts.
Data storage – the GII database: All GII data are stored, maintained and managed in the GII database. The database stores all collected data in a structured manner for all WIPO member states (not only the ranked GII economies) and for all indicators (those already included in the GII model and the new ones). It also stores data on outlier analysis (generated by the data quality checks that the GII team carries out after data collection – see below), as well as all the data queries sent to the GII data providers following an outlier analysis. As of 2024, the database will be expanded to also include country level and global aggregate data related to the Global Innovation Tracker. In addition, the micro-level data, often related to companies, used in the aggregation of certain GII indicators (e.g., Global corporate R&D investors, companies’ Unicorn valuation, companies’ Intangible asset intensity, Global brand value, etc.) has been further expanded and standardized.
The GII repository of collaborative codes: The GII repository of collaborative codes is on GitHub, which is one of the largest code-hosting platforms for version control and collaboration. The GII repository contains eight repositories in the statistical programming language R (R-codes), which are linked to diverse elements of the GII workflow and the GII report, enabling data collection, data calculation and data quality control of all GII indicators. In 2024, an updated repository for the Global Innovation Tracker – including for trends calculations at the country level, was further developed and expanded.
The GII R-package for the calculation of the GII model: The GII R-package is a custom-built package of tools, created using R, to calculate the GII model and analyze its results. The structure of the tailor-made GII R-package follows the general COINr R-package, which was developed by the European Commission Joint Research Centre (JRC) and follows the steps in the OECD/JRC Handbook for constructing composite indicators.
Assuring data quality control is at the center of the GII methodology and processes. Each collected indicator for the GII undergoes a data quality control and data audit process every year. Several data tests and analyses are performed on all collected indicators, including the analysis of means, identification of outliers based on mean and z-scores for both unscaled and scaled data, analysis of rank changes, analysis of missing data and analysis of outdated data. Following these analyses, the GII team goes back to the data providers for any necessary clarification and, when required, the data providers themselves correct the data at the source. These additional exhaustive checks ensure the reliability of all data used in the GII.
This infrastructure enables a complete workflow that links data storage and data quality control with data analysis (GII rankings and the GII report) in a fully integrated way, increasing the overall robustness of the GII data and model.
In 2024, emphasis has been given to the visualization and improved presentation of the GII data and results through the new GII Innovation Ecosystems and Data Explorer 2024. In collaboration with OneTandem, the data explorer lets users dynamically generate GII economy briefs, profiles and country comparisons seamlessly, and to look into the time series of all GII indicators, including into individual data and micro-data on intangible assets, top universities, the most valuable brands and others. In 2024, data on the Clusters Ranking, including individual Cluster briefs have been added to the website. The Data Explorer is also available for use on mobile phones.
Moving ahead, the GII team will continue exploring and improving the measurement of innovation through the GII Data Lab. By experimenting with data and novel data-driven approaches, the GII Data Lab aims to improve the measurement of innovation performance through the GII model, and to help innovation stakeholders and policymakers to make more informed decisions about innovation policy, funding, and strategy. As of 2024, the GII Data Lab focuses on three thematic research lines: (1) Innovation Finance; (2) Entrepreneurship, startups, and gazelles; and (3) Innovation impact; and a transversal line on big data and new computational methods.
Adjustments to the GII model in 2024
Appendix Table 1 summarizes the adjustments made to the GII 2024 framework. Two indicators are combined into a single indicator, creating a change in methodology. In addition, there are two new indicators and three indicators have been dropped from the framework. Due to the addition and removal of these indicators, the numbering of four remaining indicators have been adjusted, but without altering their methodology. Lastly, the name of one indicator has been modified under request of the data provider.
Data limitations and treatment
This year, the GII model includes 133 economies, which represent 92.8 percent of the world’s population and 97.5 percent of the world’s GDP in purchasing power parity current international dollars.
The timeliest possible indicators are used for the GII 2024: from the non-missing data, 2.7 percent are from 2024, 32.2 percent are from 2023, 45.8 percent are from 2022, 9.5 percent are from 2021, 3.9 percent are from 2020, 1.6 percent are from 2019 and the small remainder of 4.3 percent are from earlier years.
The GII 2024 model includes 78 indicators, which fall into three categories:
quantitative/objective/hard data (63 indicators);
composite indicators/index data (10 indicators); and
survey/qualitative/subjective/soft data (5 indicators).
This year, for an economy to feature in the GII 2024, the minimum data coverage requirement is at least 35 indicators in the Innovation Input Sub-Index (66 percent) and 16 indicators in the Innovation Output Sub-Index (66 percent), with scores for at least two sub-pillars per pillar. This year, 6.1.3 – Utility models by origin/bn PPP$ GDP has been excluded from the minimum data coverage (DMC) requirement. In the GII 2024, 133 economies had sufficient data available to be included in the Index. A total of 117 economies did not make it into the GII 2024 due to a lack of available data. For each economy, only the most recent yearly data were considered. As a rule, the GII indicators consider data from as far back as 2014.
Missing values
For the sake of transparency and replicability of results, missing values are not estimated; they are indicated with “n/a” and are not considered in the sub-pillar score. In other words, missing indicators do not translate into a zero for the country in question; the indicator is simply not taken into consideration in the aggregation process.
That said, the audit undertaken by the European Commission’s Competence Centre on Composite Indicators and Scoreboards at the Joint Research Centre (JRC-COIN) (see Appendix II) assesses the robustness of the GII modeling choices (no imputation of missing data, fixed predefined weights and arithmetic averages) by imputing missing data, applying random sets of perturbed weights and using geometric averages. Since 2012, based on this assessment, a confidence interval has been provided for each ranking in the GII as well as for the Input and Output Sub-Indices (Appendix II).
Treatment of series with outliers
Potentially problematic indicators with outliers that could polarize results and unduly bias the rankings were treated according to the rules listed below, as per the recommendations of the JRC-COIN. Only hard data indicators were treated (32 out of 63).
First rule: selection
Indicators were classified as problematic if they had:
an absolute value of skewness greater than 2.25; and
kurtosis greater than 3.5.
(5)Based on Groeneveld and Meeden (1984), which sets the criteria of absolute skewness above 1 and kurtosis above 3.5. The skewness criterion was relaxed to accommodate the small sample under consideration (133 economies).
Second rule: treatment
Indicators with between one and five outliers (27 cases) were winsorized; the values distorting the indicator distribution were assigned the next highest value, up to the level where skewness and/or kurtosis had the values specified above.
Indicators with five or more outliers, and for which skewness or kurtosis did not fall within the ranges specified above, were transformed using natural logarithms after multiplication by a given factor f.
where “min” and “max” are the minimum and maximum indicator sample values, respectively.
This formula achieves two things: it converts all series into “goods” and scales the series within the range [1, max] so that natural logs are positive, starting at 0, where “min” and “max” are the minimum and maximum indicator sample values. The corresponding formula for “bads” is:
Normalization
The 78 indicators were then normalized into the [0, 100] range, with higher scores representing better outcomes. Normalization was undertaken according to the min–max method, where the “min” and “max” values were the minimum and maximum indicator sample values, respectively. Following the recommendation of the JRC-COIN, all indicators, including index and survey data, were normalized to a 0–100 range. This normalization ensures that all indicators share the same range, facilitating their individual contribution to the overall index score.
Weights
In 2012, the JRC-COIN and GII team made a joint decision that scaling coefficients of 0.5 or 1.0 should be used instead of importance coefficients. This decision aimed to achieve balanced sub-pillar and pillar scores by considering the underlying components. In other words, the goal was to ensure that indicators and sub-pillars contribute a similar amount of variance to their respective sub-pillars/pillars.
To prevent multicollinearity during the aggregation process, any indicators within a sub-index that exhibited a high correlation, exceeding an absolute correlation of 0.95, were assigned a weight of 0.5. In 2024, there were no indicators that received a 0.5 weight, and thus all indicators had a weight of 1. Additionally, two sub-pillars – 7.2 Creative goods and services and 7.3 Online creativity – were also assigned a weight of 0.5.
Strengths and weaknesses
Strengths and weaknesses are calculated for all economies covered in the GII and are presented in the individual economy profiles (see the explanatory section Economy profiles). In simple terms, strengths and weaknesses are the top- and bottom-ranked indicators for each country. In addition, income group strengths and weaknesses are also provided, which are the respective high- and low-performing indicators within income groups.
The methodology for the calculation of strengths and weaknesses is as follows:
The scores of each indicator are converted to percentile ranks.
Strengths are defined as the indicators of an economy that have a percentile rank greater than or equal to the 10th percentile rank (across the indicators of that economy). Note that this can result in more than 10 strengths in the event of tied results.
Weaknesses are defined in an equivalent manner for the bottom 10 indicators.
If a country has an indicator that ranks equal to or lower than three, it is automatically a strength, regardless of the percentile rank.
Importantly, although the cut-off value used to define the strengths (i.e., the 10th highest percentile rank) is calculated using only indicator percentile ranks, it is also applied to sub-pillars and pillars.
In addition, for pillars and sub-pillars that do not meet the Data Minimum Coverage (DMC) criteria, strengths and weaknesses are not signaled. Pillars and sub-pillars that do not meet the DMC show the pillars and sub-pillars in brackets in the economy profiles.
Income group strengths and weaknesses are somewhat similar to overall strengths and weaknesses but are defined within income groups and use means and standard deviations. The methodology for the calculation of income group strengths and weaknesses is as follows:
For a given economy, income group strengths are those scores that are above the income group average plus the standard deviation within the group.
For that economy, weaknesses are those scores that are below the income group average minus the standard deviation within the group.
The only exceptions to the income group strengths and weaknesses are the top 25 high-income economies, where these strengths and weaknesses are computed within the top 25 group.
As the only non-high-income economy in the top 25, China’s income group strengths and weaknesses are computed within the non-top 25 group.
Since, occasionally, the low threshold for weaknesses is below zero, any score of zero is automatically marked as a weakness.
Finally, as of 2023 and following the recommendation of the audit by the WIPO Internal Oversight Section,
(8)IOD Ref: IA 2022-03, April 14, 2023: www.wipo.int/export/sites/www/about-wipo/en/oversight/docs/iaod/audit/audit-gii.pdf. strengths and weaknesses are reset, or not signaled, where the data year for a given indicator is older than the indicator mode minus five years. In practice, for the GII 2024, this means that for indicators with a data year mode of 2023, the data year of an economy must be 2018 or later to qualify as a strength or weakness.
Caveats on the year-to-year comparison of rankings
The GII compares the performance of national innovation systems across economies and presents the changes in economy rankings over time.
It is important to note that scores and rankings are not directly comparable between one year and another. Each ranking reflects the relative position of a particular economy based on the conceptual framework, the data coverage and the sample of economies of that specific GII edition, and also reflects changes in the underlying indicators at source and in data availability.
A number of factors influence the year-on-year rankings of an economy:
the actual performance of the economy in question;
adjustments made to the GII framework (changes in indicator composition and measurement revisions);
data updates, the treatment of outliers and missing values; and
the inclusion or exclusion of economies in the sample.
Additionally, the following characteristics complicate the time-series analysis based on simple GII rankings or scores:
Missing values: The GII produces relative index scores, which means that a missing value for one economy affects the index score of other economies. Because the number of missing values decreases every year, this problem reduces overtime.
Reference year: The data underlying the GII do not refer to a single year but to several years, depending on the latest available year for any given variable. In addition, the reference years for different variables are not the same for each economy, due to measures to limit the number of missing data points.
Scaling factors: Most GII variables are scaled using either GDP or population, with the intention of enabling cross-economy comparability. However, this implies that year-on-year changes in individual indicators may be driven either by the variable (numerator) or by its scaling factor (denominator).
Consistent data collection: Measuring the change in year-on-year performance relies on the consistent collection of data over time. Changes in the definition of variables or in the data collection process could create movements in the rankings that are unrelated to performance.
A detailed economy study based on the GII database and the economy profile over time, coupled with analytical work on the ground, including that of innovation actors and decision-makers, yields the best results in terms of monitoring an economy’s innovation performance, as well as identifying possible avenues for improvement.