Methodology
The SSPI is calculated for forty-nine countries, which comprise the Organisation for Economic Cooperation and Development (OECD), the Group of Twenty (G20), plus another five countries classified as ‘High income’ (2018) by the World Bank 2018. These forty-nine countries include the largest and the most industrialized economics, for which most of the policy data are available and reliable. Collectively these countries account for 90% of world GDP and approximately 67% of the world population. They also account for the large majority of greenhouse gas emissions.
Our research team relied upon a large body of literature that described and analyzed the various policies and programs, and focused on best practices across countries or in selected countries. We especially relied upon the policies supporting the Sustainable Development Goals. Although we cannot present the studies used in all the SSPI policies, some of the data sources provide references to relevant studies. Some examples of the studies we used in creating specific policy indicators are: labor market policies and how they impact labor market operations (Card and Oreopoulos 2019); policies to protect groundwater with sustainable provision of water (GlobeScan 2019); and land management policies for natural carbon sequestration in forests. (Ni et al 2016)
In selecting data to represent the policy indicators and then aggregating them into a composite index, we followed the OECD recommended practice (OECD 2008). The policy indicators were selected based on data that are well-defined and reliable, plus made publicly available over time. We only used data from credible organizations with extensive data covering many countries. Objective administrative data is selected over subjective survey data. In order to evaluate if two policy variables represented the same information, and were thus interchangeable, correlations of the variables were compared. When indicators are highly correlated and thought to represent the same information, then the indicator with more country observations or higher quality data is used. We also used sensitivity testing to evaluate how sensitive an indicator was to the use of different variables and their aggregation into the relevant category.
In designing policy indicators, we opted for the most direct measurement of policy for which reliable data were available for 2018. Where possible, direct measurements of policy were used to create policy indicators; examples include Biodiversity (the proportion of ecologically important areas protected by law), Tax Revenue (the amount of revenue raised as a percentage of GDP), and the government expenditure component of Research and Development (the proportion of GDP spent on government R&D). Direct numerical measurements of policy are possible only when governments control the level of a single parameter that determines the strength of a policy.
Many of the policies needed to achieve sustainable and shared prosperity are not so simple, however; some policy goals require a multifaceted package of policies adapted to local circumstances and constraints that work together to achieve the goal. In such situations, we use outcome metrics to proxy for these packages of government policies. For example, although Deforestation (percentage change in forest covering from a 19901999 benchmark), Fatal Injuries (fatal injuries per 10000 workers), and Primary School Net Enrollment (percentage of children of primary school age enrolled in formal education) are not specified directly by policy in the same way the amount spent on government research is, government policy nevertheless determines the levels of these variables, especially over the long run. Thus, proxy measures acknowledge the variety and plurality of specific policies, each adapted to a particular geographic, cultural, and political context, that countries may pursue to achieve a universal policy goal.
The Indicator Table describes the data used in constructing the three Pillars. Public Goods is the broadest pillar and includes government programs of the goods and services that are directly supplied by the government. What unifies the Public Goods policies is that they are largely under direct control of the national or regional government. The Public Goods pillar includes six categories: Education, Health Care, Infrastructure, Rights, Public Safety, and Global Role.
The Market Structure pillar brings together a wide array of policies that regulate and structure how markets function. The supply side is regulated through employment policies, taxation, and the protection of property and competition rights. The demand side is supported by economic security policies, with the financial system supporting the effective operation of markets. The five Market Structure categories are Employment, Economic Security, Taxation, Property, and Financial System.
The policies in the Public Goods and Market Structure pillars do not capture the responsibility that governments have to protect the environment for people today and for future generations, which explains the role of the Sustainability pillar. The Sustainability policies relate to the management and direct use of existing natural resources and the externalities resulting from degradation of the country’s natural capital. A company’s production decisions are based on its costs, which typically do not include external costs related to public health or environmental deterioration. The public ends up paying the external costs of production through worsened health or a degraded environment. Sustainability policies measure the extent to which governments have policies in place to care for the environment. The pillar includes five categories: Ecosystems, Land, Energy, Greenhouse Gases, and Waste.
The Pillars and Categories should be thought of as systems of policies that operate together, rather than as individual policies that can be analyzed separately. For example, many types of policies affect how markets operate, or the health of the environment, or the quality of life. The goal is to have policies that influence the quality of life and the environment within the overall socioeconomic system. For example, the Infrastructure category brings together a wide range of policies that together support daily activities in a community and also impact the environment. Infrastructure policies include access to electricity, water quality, basic sanitation, internet access, and transportation network, with the infrastructure commonly built by governments and service either directly provided by the government or regulated through private companies.
All indicators are set up so that a higher score represents a policy oriented toward a superior outcome. Each policy indicator is normalized to a value between 0 and 1, so that the widely varying data are expressed in comparable units across all indicators. Indicators are normalized based upon lower and upper goalposts, which represent the hypothetical minimum and hypothetical maximum for the policy variable, using the following
Our steps for calculating scores for each indicator aligns with the same methodology of goalposting for the Human Development Index. The HDR 2020 technical explains how these “dimension indices” (our equivalent of indicator scores) are normalized to values between 0 and 1, and these minimum and maximum goalposts are set according to both historical evidence and achievability.
Setting the goalposts of a specific indicator shifts the center and controls the spread of its score distribution while preserving the ranks in the observed data. When possible, we used internationally established norms as goalposts for policies, such as those in the 2022 Sustainable Development Report (Sachs et al 2022). For policy indicators without an established norm, the historical values observed across countries are used as a guide for what is possible for high performers while allowing for continued improvement in future years, and also to benchmark low performance. We used sensitivity testing of the goalposts of a specific indicator to evaluate the distribution and outliers across countries, and used the observed high values as a benchmark of what good policy can achieve as the upper goalpost. A few very high scores were determined to reflect noise, and the scores were capped at the upper goalpost. Setting the lower goalpost too low can penalize countries with insufficient resources or characteristics, such as geography or culture, that may constrain policy. Therefore some of our selected lower goalposts cap the observed values at the lower goalpost, both to minimize noise and to reflect a reasonable lower benchmark for weak country policies.
For example, for the policy to protect biodiversity, benchmarks can be drawn from environmental science. We know that to protect the planet’s ecosystems, nations must quit causing extinction and preserve species to maintain their biodiversity. The percentage of important sites covered by protected areas ranges from 0 to 100, with the range given as a percentage of internationally known sites. The maximum observed value is 96.9%, so it is reasonable to expect a maximum of 100%, since ideally we would want a country’s policy to protect known important sites.
A more complex example is provided by creating goalposts for policies relating to carbon sequestration. We do not know the maximum amount of carbon that could be sequestered because it varies widely by country and relevant data across countries are sparse. Without clear benchmarks across countries, we rely upon the data, including country land characteristics and historical trends, to guide us in setting goalposts for what a good policy can achieve.
Historical trends of the percentage change in how land is allocated for a particular purpose provides an observable benchmark over time of how well the land use management is doing in achieving specific goals. In the case of carbon sequestration, we rank country policy based on carbon density, i.e., the ratio of carbon stock in living biomass over forest land (hectares) in 2018 compared to the 1990s average, when deforestation became noticeable. Higher (positive) values indicate that a country is prioritizing and allocating land to forests that are sequestering carbon. This method uses the country’s existing land characteristics for the benchmark, rather than using a universal benchmark for how good a country’s land use should be. The indicator for deforestation follows a similar rationale, and we use data on how well the country has maintained or increased forest coverage over time: the ratio of naturally regenerating forests in 2018 compared to the 1990s average.
Both extremely high and low outliers are censored and assigned the lower or upper goalpost value. Our approach is similar to the approach used by the SDG Index in setting upper targets based on known goals or science, and otherwise based on the average of top performers; and in setting lower targets at the 2.5th percentile of the distribution. Each SDG indicator distribution is censored, with values exceeding the upper bound scored 100, and values below the lower bound scored 0.14.