Appendix: Methodology of the Inclusive Growth and Development Benchmarking Framework
The approach presented in this Report is intended to be normative and primarily aimed at stimulating discussion on policy priorities, actions that could be taken by the private sector (alone or in concert with government), and further research endeavors. As outlined above, there is widespread agreement that the growth process must yield more inclusive outcomes, and research on the factors that determine such outcomes is ongoing and remains at a formative stage. Many determinants are thought to influence the process and benefits of growth outcomes and the way in which they are distributed. The selection of the pillars therefore represents a key assumption of the Framework. It is grounded in available research and best judgment based on historical experience. However, these domains have not yet been empirically proven to have a direct, causal link to increased growth or social equity, either individually or collectively.
For practical reasons, the framework separates prioritized policy domains into seven distinct pillars, as though these are interdependent and interconnected – they tend to reinforce each other, and a weakness in one area often has a negative impact on others. No single determinant can ensure inclusive growth, which can only be achieved through a combination of factors. For example, employment can only contribute to equitable growth if education is widely accessible and transmits skills of relevance to the labor market. Private-sector investment will be higher and more efficient if government and business activity is transparent and ethical. Likewise, education is also linked to health outcomes – in advanced economies, those with the highest education can expect to live six years longer than their poorly educated peers.
The appropriate mix of policies and institutions will depend on country circumstances and preferences, so the Framework does not include an overall aggregate ranking or league table of countries. Similarly, it does not intend to suggest that there is an ideal policy or institutional mix for the pursuit of inclusive growth and development that will apply to all countries. For the same reason, the Framework does not assign different weights to the pillars and sub-pillars.
Given the data limitations, the complexity of the topic, and the need for further research, the individual indicators should be interpreted as simple proxies for prevailing conditions and the extent to which countries are utilizing their policy space. A weak or strong score should thus be seen as a marker or signpost of where a country might explore policy changes or other actions.
It is important to note that in a number of instances, data had to be adjusted to take into account both equity and growth considerations. Although equity remains a principal focus when assigning rank direction, a cut-off sometimes has been applied at the point where these policies might dampen growth. These trade-offs are present in the case of some labor and tax-related indicators, where a particularly high degree of protection or taxation can begin to dampen growth. Other adjustments were undertaken if the relationship between the indicator and inclusive growth is not linear. For example, paid maternity leave is beneficial to female inclusion until it begins to adversely affect wages and (re)integration into the labor market. Similarly, some financial market indicators, such as domestic credit to the private sector or share turnover, can are characterized by negative effects at both extremes. Specific thresholds have been set were based upon available literature and the authors’ interpretation of the data.
Data and Aggregation Methods
The Country Profiles include two types of data. The first category is quantitative data collected from leading international organizations and other respected sources. The second category of data is derived from the World Economic Forum’s Executive Opinion Survey, which assesses the perspectives of more than 14,000 business leaders about their countries’ business and political environment (between February and June 2014). The responses from the survey are on a 1-to-7 scale, with 1 representing the worst case, and 7 the best.
If quantitative data presents outliers, data thresholds are introduced to reduce the bias in the distribution of the data. The same thresholds are applied across the full sample of countries where data is available to allow for some degree of comparability (at indicator level and across some sub-pillars).
The computation is based on successive aggregations of scores from the indicator level to the sub-pillar and pillar level. Unless noted otherwise, an arithmetic mean is used to aggregate individual indicators within a category. For quantitative data, to make aggregation possible, indicators are converted to a 1-to-7 scale (worst to best) in order to align them with the Survey results. A linear min-max transformation is applied, which preserves the order of, and the relative distance between, country scores.
a. Formally, for a category i composed of K indicators, there is:
b. Formally, the equation is:
The sample minimum and sample maximum are, respectively, the lowest and highest country scores in the sample of economies covered by the benchmarking tool. In some instances, adjustments were made to account for extreme outliers. For those indicators for which a higher value indicates a worse outcome, the transformation formula takes the following form, thus ensuring that 1 and 7 still correspond to the worst and best possible outcomes, respectively:
In order to facilitate peer-group comparisons for countries, the results are grouped into the four broad categories of countries based on a combination of the World Economic Forum’s Global Competitiveness Index methodology and the World Bank’s 2015 income classifications that were available at the time the Report was drafted: advanced, upper-middle, lower-middle and low income.1 This classification also reflects somewhat different available data sets and policy challenges for each group. The income thresholds presented in the table below are based on GDP per capita in current US dollars.
Results are displayed by pillar as well as by country (scorecards). The former is intended to enable the reader to benchmark a given score against a peer group of countries in a given policy domain and across other policy domains. The latter is intended to provide a comprehensive picture of a country’s performance and enabling environment conditions across the full spectrum of policy domains covered by the Benchmarking Framework. In addition to numerical values, a five-color system of color shading is applied to ease interpretation of the data and comparisons across countries and indicators, with darkest green representing the best performance in a pillar, shades of yellow standing for average performance, and deepest red displaying the poorest performance. The same color palette has been used for the icons on the country profiles showing the individual country performances as well as in the aggregated pillar result tables for each income group. This allows both an internal comparison for individual countries (by showing in which pillars they perform more or less well) as well as a cross-country comparison (how the countries compare to their peers in the various pillars and sub-pillars).
It is important to note that in order to facilitate the comparison of countries with their peers – those with similar resources at their disposal – the color palette has been based on results by income group. Thus, caution must be taken in comparing color results across income groups, as they are not directly comparable. Specifically, the range of colors shown for advanced and upper-middle income economies are each based on the results of the specific income group and only comparable to the countries within their group. For lower-middle income and low-income countries, a single color calibration has been performed based on the range in scores of the lower-middle income countries. This has been done to highlight the still significant room for improvement even for the best performers within the low income group.2
The Report covers 112 countries representing all regions. Country coverage has mainly been driven by data availability – all but 24 countries have full coverage on all pillars, and no countries have more than a third of missing data in a given pillar.3 In most cases, missing values do not exceed 25%. If the overall results of more than two pillars could not be properly calculated, the country has not been included. The Forum will strive to expand coverage as more comparable data becomes available, especially for low income countries. For this reason, for some variables two distinct data sets have been used (one for advanced and upper-middle income economies and another for lower-middle income and low income economies) in order to capture a wide array of concepts and to use the best data available for a large range of countries. For example, for advanced and upper-middle income countries, data from the OECD’s PISA assessment has been included, while for lower-middle income and low income countries UNESCO’s WIDE Database on Educational Inequality has been used due to the lack of comparable data by income quintile across the whole sample. This is also the case for a few other indicators that are available for higher income economies but not available for some of the other country groupings. As a result, pillar level scores are not strictly comparable between income groups. The table below indicates the specific variables that are available only for certain income groups.
Strengthening the World Economic Forum’s Framework for Inclusive Growth
Some key concepts that are important for inclusive growth could not be captured due to gaps in available data – for example, discrimination against the disabled, migrants, and ethnic minorities. Data is especially scarce for low income countries and capturing the distribution of outcomes by income groups. Going forward, in order to make progress in this area, countries and international organizations will need to regularly collect better data in these critical areas especially through the use of household surveys. It is very hard to fix what you cannot measure.
It bears mention that measures of real economy investment, or productive uses of capital, are a relatively underexplored area with important implications for inclusive growth. For this pillar, comparable data for a large number of countries is limited, necessitating the use of several different variables or proxies in order to capture this complex concept. For example, it is difficult to capture net equity issuance (taking into account share buybacks) in a single measure due to poor country coverage; these indicators could not be combined and have been presented separately in this Report. Likewise, private investment in infrastructure data is only available for developing countries as data for many advanced economies also includes public investment. The Forum’s goal is to provide a more complete breakdown of this concept in the next Report.
This Report should be seen as marking the start of an ongoing process. Empirical research on the topic of inclusive growth is still emerging. As it evolves, the Forum intends to use it to explore the relationships and relative importance of the different pillars. Work will also be done to incorporate new countries and indicators into the analysis and to test the robustness of the Framework. This work on further refining and upgrading the methodology will inform the next edition of the Report.