Risk dimensions and data sources

2. Risk Dimensions and data sources

The risk dimensions listed in the tool have been identified by SIFAV members as being the most important. These risk dimensions should be viewed as a first iteration, others can be added over time if users are specifically requesting this. 

For each of the risk dimensions, consultancy Blue North Sustainability assessed the available data sources. Based on this analysis a credible source for each of the risk dimensions was identified. These were compiled in one comprehensive database which will be updated once a year. 

The Risk Dimensions, grouped by Risk Categories, are presented in the summary table below, showing the recommended Data Source for each. Three new Risk Dimensions were added in 2023, they are: 

• Discrimination;
• Indigenous & Community Land Rights; 
• Ozone-Depleting Substances;

Risk Category	Risk Dimension	Data Source
Social	Child Labour	UNICEF  US Department of Labour
	Discrimination	Fundamental Rights and Civil rights factors of the World Justice Project Rule of Law Index
	Forced & Bonded Labour	Global Slavery Index
	Freedom of Association	ITUC Global Rights Index
	Healthy & Safe Workplace	ILOSTAT
	Migrant Labour	ILOSTAT
	Working Poverty	ILOSTAT
	Women’s Rights & Gender Equality	UNDP Human Development Index
	Working Hours	ILOSTAT
	Indigenous & Community Land Rights	The Global Platform of Indigenous and Community Lands
Environmental	Climate Change Vulnerability	Germanwatch Global Climate Risk Index
	GHG Emissions	Blonk Sustainability
	Food Loss & Waste	FAOSTAT
	Pesticide use	FAOSTAT
	Fertilizer use	Environmental Performance Index
	Biodiversity	Resource Watch
	Deforestation	FAO Global Forest Resources Assessment
	Soil Degradation	European Soil Data Centre
	Overall Water Risk	WWF Water Risk Filter
	Ozone-Depleting Substances	UN Environment Programme
Governance & other	Institutional Arrangements / Good Governance	World Bank Worldwide Governance Indicators

2.1 Child labor (social)

Risk Definition: The term “child labor” is often defined as work that deprives children of their childhood, their potential, and their dignity, and that is harmful to physical and mental development. It refers to work that: is mentally, physically, socially, or morally dangerous and harmful to children; and/or interferes with their schooling by depriving them of the opportunity to attend school; obliging them to leave school prematurely; or requiring them to attempt to combine school attendance with excessively long and heavy work.

Data source: UNICEF Child Labour Statistics (“Percentage of children aged 5-17 years engaged in child labor (by sex)”) (link to the dataset) and The US Department of Labor's Annual Findings on the Worst Forms of Child Labor (link to the report).

Data attributes:

• Data level: Country
• Data scope: Generic (not product specific)
• Source update frequency: Annually

Description of the data: UNICEF compiles child labor statistics to make internationally comparable datasets publicly available. The data measures the proportion and number of children engaged in child labor per country. The number of children engaged in child labor corresponds to the number of children reported to be in child labor during the reference period. The proportion of children in child labor is calculated as the number of children in child labor divided by the total number of children in the population. For this indicator, children include all persons 5 to 17 years of age. The UNICEF dataset is the primary source because it is regularly updated, and the organisation focuses on children. Missing data is supplemented with the US Department of Labour report.

• UNICEF covers 96 countries and considers children aged 5 to 17. It was last updated in May 2022.
• The US Department of Labor publishes a comprehensive annual report covering 131 countries and considering children aged 5 to 14. The latest report is from 2021.

Notes on the dataset: The data for this dimension was transformed into risk scores from 0 to 10 using two steps: windsorisation and normalization. Windsorisation is a technique that replaces extreme values with more moderate ones, to reduce the impact of outliers on the analysis. Normalization is a technique that rescales the data to a common range, such as 0 to 10, to make it comparable across different units or scales. By applying these two steps, we obtained risk scores that are more robust and consistent for this dimension.

2.2 Discrimination

Risk Definition: Discrimination is the unjust or prejudicial treatment of different categories of people, especially based on race, age, sex, or disability.

Data source: The World Justice Project (WJP) is an independent, multidisciplinary organisation working to create knowledge, build awareness, and stimulate action to advance the rule of law worldwide.

Data Attributes:

• Data level: Country
• Data scope: Generic (not product specific)
• Source update frequency: Annually

Description of the data: The 2022 WJP Rule of Law Index evaluates 140 countries and jurisdictions worldwide. The index considers various issues/factors. Two apply to the discrimination indicator, namely:

• Fundamental Rights: Equal treatment and absence of discrimination (Factor 4.1), which measures whether individuals are free from discrimination—based on socio-economic status, gender, ethnicity, religion, national origin, sexual orientation, or gender identity - with respect to public services, employment, court proceedings, and the justice system; and
• Civil Justice: Civil Justice is free of discrimination (Factor 7.2) which measures whether the civil justice system discriminates in practice based on socioeconomic status, gender, ethnicity, religion, national origin, sexual orientation, or gender identity.

The index score ranges from 0 (weaker adherence to the rule of law) to 1 (stronger adherence to the rule of law).

Notes on the dataset: The data for this dimension was transformed into risk scores from 0 to 10 using two steps: windsorisation and normalization. Windsorisation is a technique that replaces extreme values with more moderate ones, to reduce the impact of outliers on the analysis. Normalization is a technique that rescales the data to a common range, such as 0 to 10, to make it comparable across different units or scales. By applying these two steps, we obtained risk scores that are more robust and consistent for this dimension.

2.3 Forced & Bonded Labour (social)

Risk Definition: Forced labor is any work or service which people are forced to do against their will, under threat of punishment. Bonded labor, also known as debt bondage and peonage, occurs when people give themselves into slavery as security against a loan or when they inherit a debt from a relative. 

Data source: Global Slavery Index (link to the dataset). Walk Free’s Global Slavery Index has developed world-leading research to provide a measurement of the size and scale of modern slavery, as well as assess country-level vulnerability and governmental responses to the occurrence of forced and bonded labor practices. Together with the International Labour Organisation (ILO) and the International Organisation for Migration (IOM), Walk Free has developed the joint Global Estimates of Modern Slavery. 

Data attributes: 

• Data level: Country
• Data scope: Generic (not product specific)
• Source update frequency: Biannually 

Description of the data: The data is the estimated prevalence of the population in modern slavery (victims per 1,000 population). This can be ranked from low to high prevalence, therefore, producing a Prevalence Index Ranking. The underlying data used in the analysis is drawn from nationally representative surveys implemented through the Gallup World Poll, including a module on modern slavery in 48 countries, and data from the Global Slavery Index Vulnerability Model. 

Notes on the dataset: The data is from a 2018 report. This dataset, therefore, does not follow standard scoring mechanisms. To ensure alignment, the following scoring method was used:

Datapoint value [Victims per 1000 population]	Score
Less than 9	Use datapoint value as is
Between 9 and 23	9
More than 23	10

2.4 Freedom of Association (social)

Risk Definition: Freedom of association is a fundamental human right proclaimed in the Universal Declaration of Human Rights. It is the enabling right to allow effective participation of non-state actors in economic and social policy, lying at the heart of democracy and the rule of law. Ensuring that workers and employers have a voice and are represented is, therefore, essential for the effective functioning not only of labor markets but also of overall governance structures in a country 

Data source: ITUC Global Rights Index (link to the dataset). The Global Rights Index is a frequently updated index that provides a reliable rating regarding the freedom of associations at the country level.

Data attributes: 

• Data level: Country
• Data scope: Generic (not product specific)
• Source update frequency: Annually

Description of the data: This index is grounded in standards of fundamental rights at work, based on international human rights law, and in particular ILO Conventions Nos. 87 and 98, as well as the jurisprudence developed by the ILO supervisory mechanisms. The indicators that contribute to this index fall under the following categories: 

• Civil liberties 
• Right to establish and join unions 
• Trade union activities 
• Right to collective bargaining 
• Right to strike 

This index is a scoring system that classifies countries into the following five scores: 

1. Sporadic violations of rights

2. Repeated violations of rights

3. Regular violations or rights

4. Systematic violations of rights

5. No guarantee of rights

5+. No guarantee of rights due to the breakdown of the rule of law

 

The data is collected through a questionnaire sent to 331 national unions in 163 countries to report violations of workers’ rights by indicating relevant details. Regional meetings with human and trade union rights experts are held where the questionnaire is disseminated, explained, and completed. Additionally, legal researchers analyze national legislation and identify sections that are not adequately protecting internationally recognized collective labor rights. 

Notes on the dataset: The data for this dimension was transformed into risk scores from 0 to 10 using two steps: windsorization and normalization. Windsorization is a technique that replaces extreme values with more moderate ones, to reduce the impact of outliers on the analysis. Normalization is a technique that rescales the data to a common range, such as 0 to 10, to make it comparable across different units or scales. By applying these two steps, we obtained risk scores that are more robust and consistent for this dimension.

2.5 Healthy & Safe Workplace (social)

Risk Definition: Workplace health and safety is generally defined as the science of the anticipation, recognition, evaluation, and control of hazards arising in or from the workplace that could impair the health and well-being of workers. Health and safety focus on the Promotion and maintenance of the highest degree of physical, mental, and social well-being of workers in all occupations; Prevention of worker absence due to poor health caused by their working conditions; Protection of workers in their employment from risks resulting from factors adverse to health, and; Assessment of an employee’s occupational environment and adapting to their physiological and psychological capabilities.

Data source: ILOSTAT (link to fatal and link to non-fatal occupational injuries datasets). Non-fatal and fatal occupational injuries per 100'000 workers are proposed as a good indicator to assess workplace health and safety risks. It provides data on the average number of new cases of non-fatal and fatal occupational injury during the calendar year per 100,000 workers.

Data Attributes:

• Data level: Country
• Data scope: Generic (not product specific)
• Source update frequency: Annually

Notes on the dataset: The database covers 120 countries out of the world's total of 195 countries. This dataset was the most comprehensible available; however, the metrics are based on the most recent data for each country, which may create inconsistencies in the time period between countries. As far as possible, data is based on the agricultural sector. The reporting quality and details among countries vary.
The data for this dimension was transformed into risk scores from 0 to 10 using two steps: windsorisation and normalization. Windsorisation is a technique that replaces extreme values with more moderate ones, to reduce the impact of outliers on the analysis. Normalization is a technique that rescales the data to a common range, such as 0 to 10, to make it comparable across different units or scales. By applying these two steps, we obtained risk scores that are more robust and consistent for this dimension. 

2.6 Migrant labor (social)

Risk Definition: Migrant workers can be at high risk of exposure to workplace hazards and face additional work-related risk factors and unfavorable social determinants of health including employment and wage discrimination, poor working and living conditions, lack of access to social protection, and language and culture barriers. These work-related risks can result in a higher incidence of occupational injuries and work-related diseases among migrant workers, compared with non-migrant workers.

Data source: ILOSTAT (link to the dataset). There are limited global datasets reporting on migrant labor The primary data is collected by International Labour Migration Statistics (ILMS) and distinguishes between native- and foreign-born labor in each country. 

Data attributes: 

• Data level: Country
• Data scope: Generic (not product specific)
• Source update frequency: Annually

Description of the data: This annual dataset reports a country’s labor force by sex, age, and place of birth (native vs foreign-born). The labor force comprises all persons of working age who furnish the supply of labor to produce goods and services during a specified time-reference period. 

Notes on the dataset: This dataset was the most comprehensive available; however, the metrics are based on the most recent data for each country, which may create inconsistency in the time periods between countries. The data for this dimension was transformed into risk scores from 0 to 10 using two steps: windsorization and normalization. Windsorization is a technique that replaces extreme values with more moderate ones, to reduce the impact of outliers on the analysis. Normalization is a technique that rescales the data to a common range, such as 0 to 10, to make it comparable across different units or scales. By applying these two steps, we obtained risk scores that are more robust and consistent for this dimension.

2.7 Working Poverty (social)

Risk Definition: The working poverty rate conveys the percentage of employed persons living in poverty despite being employed. The international poverty line is a threshold used to measure extreme poverty based on consumption or income levels. A person is considered extremely poor if their consumption or income level falls below the minimum level necessary to meet basic needs.

Data source: ILOSTAT’s working poverty rate dataset (link to the dataset). 

Data attributes: 

• Data level: Country
• Data scope: Generic (not product specific)
• Source update frequency: Annually

Description of the data: The dataset records the Sustainable Development Goals (SDG) Indicator 1.1.1 – Working poverty rate (percentage of employed living below US$1.90 PPP). Poverty is defined using the international poverty line of US$1.90 per day in purchasing power parity (PPP). The ($1.90 a day) poverty line or the critical threshold value, below which an individual is determined to be extremely poor, allows for comparing and aggregating progress across countries in reducing the number of people living under extreme poverty and monitoring trends at the global level. In addition, poverty measures based on an international poverty line attempt to hold the real value of the poverty line constant over time, allowing for accurate assessments of progress toward meeting the goal of eradicating extreme poverty and hunger.

Notes on the dataset: The database covers 120 countries out of the world's total of 195 countries. The data for this dimension was transformed into risk scores from 0 to 10 using two steps: windsorization and normalization. Windsorization is a technique that replaces extreme values with more moderate ones, to reduce the impact of outliers on the analysis. Normalization is a technique that rescales the data to a common range, such as 0 to 10, to make it comparable across different units or scales. By applying these two steps, we obtained risk scores that are more robust and consistent for this dimension. 

2.8 Women’s Rights & Gender Equality (social)

Risk Definition: Women’s rights are the fundamental human rights that were enshrined by the United Nations for every human being on the planet nearly 70 years ago. These rights include the right to live free from violence, slavery, and discrimination; to be educated; to own property; to vote; and to earn a fair and equal wage.

Data source: The UNDP Gender Inequality Index (link to the dataset). The Gender Inequality Index derived from the Human Development Index is a reliable source to gain insight into women’s rights and gender equality in different countries. It encompasses multiple aspects of this risk and is updated regularly.

Data attributes: 

• Data level: Country
• Data scope: Generic (not product specific)
• Source update frequency: Annually

Description of the data: The Gender Inequality Index is a composite measure reflecting inequality in achievement between women and men in the following three dimensions: 

• Reproductive health 
• Empowerment 
• The labor markets. 

This Index ranges from 0 (where women and men are equal) to 1 (where one gender fares poorly as possible in all measured dimensions). This index was calculated for all countries with data of reasonable quality.

Notes on the dataset: The data for this dimension was transformed into risk scores from 0 to 10 using two steps: windsorzsation and normalization. Windsorization is a technique that replaces extreme values with more moderate ones, to reduce the impact of outliers on the analysis. Normalization is a technique that rescales the data to a common range, such as 0 to 10, to make it comparable across different units or scales. By applying these two steps, we obtained risk scores that are more robust and consistent for this dimension.

2.9 Working Hours (social)

Risk Definition: Some of the current challenges in working time are the same as those motivating the adoption of the ILO's first Convention, the Hours of Work (Industry) Convention, 1919 (No. 1): excessive hours of work and inadequate periods of rest and recuperation, which can damage workers' health and increase the risk of work accidents. In many parts of the world, there is a significant link between low wages and excessive working time. Long working hours prevent workers from getting adequate rest, attending to family responsibilities, and participating in the community. 

Data source: ILOSTAT (link to the dataset).  

Data attributes: 

• Data level: Country
• Data scope: Generic (not product specific)
• Source update frequency: Annually

Description of the data: The dataset provides the mean weekly hours worked per employee by economic activity. Data on hours of work are presented, whenever possible, on the basis of the mean number of hours of work per week, and with reference to hours worked in all jobs of employed persons and in all types of working time arrangements (e.g., full-time, and part-time). Data are disaggregated by economic activity according to the latest version of the International Standard Industrial Classification of All Economic Activities (ISIC) available for that year. 

Notes on the dataset: The dataset includes 121 countries. The economic activity "Crop and animal production, hunting, and related service activities", both male, and female, and the latest period for a particular country, are used. The data for this dimension was transformed into risk scores from 0 to 10 using two steps: Windsorization and normalization. Windsorization is a technique that replaces extreme values with more moderate ones, to reduce the impact of outliers on the analysis. Normalization is a technique that rescales the data to a common range, such as 0 to 10, to make it comparable across different units or scales. By applying these two steps, we obtained risk scores that are more robust and consistent for this dimension.

2.10 Indigenous & Community Lands Rights

Risk definition: The dataset "Indicators of the Legal Security of Indigenous and Community Lands" is characterized by a series of ten indicator questions that point to the security of the land tenure for Indigenous Peoples or communities as established in national laws. These indicators consider how the law addresses the rights of indigenous people or community lands.

Data source: "Legal Security of Indigenous and Community Lands - Indigenous Peoples" provided by the Global Platform of Indigenous and Community Lands (link to the dataset).

Data attributes: 

 

• Data level: Country
• Data scope: Generic (not product specific)
• Source update frequency: Annually

Description of the data: The average score for the ten indicators of the legal security of indigenous and community lands is used. The dataset includes 119 countries. The score provides a snapshot of the security of indigenous and community land. It does not represent an index because the various indicators are not weighted based on their relative importance to secure tenure. The scoring of indicators is based exclusively on express legal provisions.

Notes on the dataset: The data for this dimension was transformed into risk scores from 0 to 10 using two steps: windsorization and normalization. Windsorization is a technique that replaces extreme values with more moderate ones, to reduce the impact of outliers on the analysis. Normalization is a technique that rescales the data to a common range, such as 0 to 10, to make it comparable across different units or scales. By applying these two steps, we obtained risk scores that are more robust and consistent for this dimension.

2.11 Climate Change Vulnerability (environmental)

Risk definition: Climate change vulnerability was defined by the IPCC as "the degree to which a system is susceptible to and unable to cope with, adverse effects of climate change, including climate variability and extremes.” At a country level, it is a measure of the degree to which a country is exposed to climate change-related “shocks” such as extreme weather events, and its capacity to adapt and/or mitigate these shocks.

Data source: Germanwatch Global Climate Risk Index (link to the dataset). The Germanwatch Global Climate Risk Index (CRI) is an analysis based on one of the most reliable data sets available assessing the impacts of extreme weather events and associated socio-economic data, the MunichRe NatCatSERVICE. The CRI analysis ranks to what extent countries and regions have been affected by impacts of climate change related to extreme weather events (heatwaves, floods, storms, drought, etc.). 180 countries were analyzed for the 2021 CRI, and data from 2000 to 2019 is considered in this database. 

Data attributes: 

 

• Data level: Country
• Data scope: Generic (not product specific)
• Source update frequency: Annually

Description of the data: To calculate the CRI score, the following indicators, with regard to climate change-related extreme weather events, are analyzed per country: 

1. number of deaths, 

2. number of deaths per 100 000 inhabitants, 

3. sum of losses in US$ in purchasing power parity (PPP) as well as 

4. losses per unit of gross domestic product (GDP). 

Each country's index score is then derived from a country's average ranking in all four indicating categories, according to the following weighting: 

• death toll, 1/6; 
• deaths per 100 000 inhabitants,1/3; 
• absolute losses in PPP, 1/6; and 
• losses per GDP unit, 1/3. 

The lower the score, the more affected the country is and the more vulnerable it is to climate change shocks.

Notes on the dataset: The data from this database was normalized to fit the 0-10 scale used in the Sustainability Risk Assessment.

2.12 Green House Gas (GHG) Emissions (environmental)

Risk definition: Greenhouse gas (GHG) emissions from human activities strengthen the greenhouse effect, driving anthropogenic climate change. Achieving net zero or net negative GHG emissions is an important objective of ensuring sustainable societies and sustainable agriculture & food systems specifically.

Data source: Blonk Sustainability (https://www.agri-footprint.com/).  

Data attributes: 

• Data level: Country
• Data scope: Product specific
• Source update frequency: Regularly 

Description of the data: Blonk processed the data for the requested crops and countries from a kg CO2_e per kg crop value into a risk score (0 for low risk and 10 for high risk).

Notes on the dataset: The data for this dimension was transformed into risk scores from 0 to 10 using two steps: windsorization and normalization. Windsorisation is a technique that replaces extreme values with more moderate ones, to reduce the impact of outliers on the analysis. Normalization is a technique that rescales the data to a common range, such as 0 to 10, to make it comparable across different units or scales. By applying these two steps, we obtained risk scores that are more robust and consistent for this dimension.

2.13 Food loss and waste (environmental)

Risk definition: Food waste is any food, and inedible parts of food, removed from the food supply chain to be recovered or disposed of (including composted, crops plowed in/not harvested, anaerobic digestion, bio-energy production, co-generation, incineration, disposal to sewer, landfill or discarded to the sea. 

Data source: FAOSTAT Food Balances (2010-) (link to the dataset). The FAOSTAT (Food and Agriculture Organisation of the United Nations statistical database) provides access to data at a commodity and country level.

Data attributes: 

Data attributes: 

• Data level: Country
• Data scope: Product specific
• Source update frequency: Annually, the latest version is 2019 data 

Description of the data: Food losses are shown per country in 1000 metric tonnes per commodity. Not all commodities are uniquely measured in the dataset, but the dataset does include “Vegetables, other” and “Fruit, other.” These two “other” categories will be used to represent commodities not yet included in the dataset. The data set is updated and released by FAO on an annual basis. 

Notes on the dataset: The data for this dimension was transformed into risk scores from 0 to 10 using two steps: windsorization and normalization. Windsorization is a technique that replaces extreme values with more moderate ones, to reduce the impact of outliers on the analysis. Normalization is a technique that rescales the data to a common range, such as 0 to 10, to make it comparable across different units or scales. By applying these two steps, we obtained risk scores that are more robust and consistent for this dimension.

2.14 Pesticide use (environmental)

Risk definition: The significant volume of chemical contaminants currently found in the air, soil, water, and sediment can have detrimental implications for the environment. Most of the chemical contamination is a result of human activities. Pesticides are chemicals that are intentionally added to the environment to control crop pests and diseases. Pesticides that potentially result in adverse biological effects – whether at an individual, population, community, or ecosystem level, are of particular concern in agricultural systems. 

Data source: FAOSTAT Pesticide Use (link to the dataset) and FAOSTAT Land Use > Cropland (link to the dataset). The land use > cropland dataset is used to calculate the pesticide use intensity in kg pesticide per hectare cropland. 

Data attributes: 

• Data level: Country
• Data scope: Generic (not product specific)
• Source update frequency: Annually

Description of the data: The total pesticide use data (1990 to 2019) includes insecticides, fungicides, bactericides, herbicides, plant growth regulators, rodenticides, mineral oils, disinfectants, and others.

Notes on the dataset: The data for this dimension was transformed into risk scores from 0 to 10 using two steps: windsorization and normalization. Windsorization is a technique that replaces extreme values with more moderate ones, to reduce the impact of outliers on the analysis. Normalization is a technique that rescales the data to a common range, such as 0 to 10, to make it comparable across different units or scales. By applying these two steps, we obtained risk scores that are more robust and consistent for this dimension.

 

2.15 Fertilizer use (environmental)

Risk definition: Fertilisers rich in nitrogen are intentionally added to the environment for the provision of nutrients. It supports plant growth and is vital to the agricultural sector; however, when mismanaged, fertilizers can cause widespread damage through nitrogen pollution. Fertilizers can lead to adverse biological effects – whether at an individual, population, community, or ecosystem level.

Data source: The Environmental Performance Index (EPI) and Sustainable Nitrogen Management Index (SNMI) (link to the dataset). The EPI database provides a risk rating, which can be viewed online or downloaded in pdf format. It is a credible resource. For fertilizer use, the database covers all countries; however, it is not commodity specific.

Data attributes:

• Data level: Country
• Data scope: Generic (not product specific)
• Source update frequency: Annually

Description of the data: The SNMI seeks to balance the efficient application of nitrogen fertiliser with maximum crop yields as a measure of the environmental performance of agricultural production. First, countries are assessed by their nitrogen use efficiency (NUE), the ratio of nitrogen absorbed by harvested crops during growth to the nitrogen inputs, including fertilizer. Second, countries are assessed on annual nitrogen yield, which is the amount of nitrogen bound up in harvested crops every year. A risk score of 100 indicates that a country is optimizing both crop yields and fertilizer application, and a score of 0 indicates a country has the worst performance on the SNMI. Data supporting this metric comes from FAOSTAT and covers 197 countries since 1961.

Notes on the dataset: The FAOSTAT database provides historical records of nitrogen fertilizer use but does not provide a breakdown of how the fertilizers have been used for pastures versus different crop types or soil types. The data for this dimension was transformed into risk scores from 0 to 10 using two steps: windsorization and normalization. Windsorization is a technique that replaces extreme values with more moderate ones, to reduce the impact of outliers on the analysis. Normalization is a technique that rescales the data to a common range, such as 0 to 10, to make it comparable across different units or scales. By applying these two steps, we obtained risk scores that are more robust and consistent for this dimension. 

2.16 Biodiversity (environmental)

Risk definition: Biodiversity refers to the variety of plant and animal life in the world or a particular habitat, a high level of which is usually considered important and desirable. Causes of decreased biodiversity include the spread of invasive alien species, pollution, climate change, and habitat loss due to land use change. 

Data source: Resource Watch (link to the dataset) Provides Biodiversity Intactness Index data (discussed in more detail below).  

Data attributes: 

• Data level: Regional
• Data scope: Generic (not product specific)
• Source update frequency: Based on 2005 data with annual projections 

Description of the data: The Biodiversity Intactness Index (BII) is a widely recognized measure of biodiversity used in key research reporting on the state of biodiversity such as the WWF Living Planet Report and the IPCC Report. The BII covers over 54 000 species including birds, mammals, plants, fungi, and insects. The BII measures the proportion of natural biodiversity remaining in local ecosystems relative to an intact ecosystem undisturbed by human activity. The index score is represented as a percentage. Scores less than 60% are considered beyond the ‘safe’ boundary. In contrast, scores higher than 95% indicate that biodiversity has been relatively unimpacted by human activity in the area. The datasets used 2,382,624 data points from the Projecting Responses of Ecological Diversity in Changing Terrestrial Systems (PREDICTS) database, a global database of how local terrestrial biodiversity responds to human impacts. The models included 4 pressure variables—land use, land use intensity, human population density, and proximity to the nearest road.

Notes on the dataset: The data for this dimension was transformed into risk scores from 0 to 10 using two steps: windsorization and normalization. Windsorization is a technique that replaces extreme values with more moderate ones, to reduce the impact of outliers on the analysis. Normalization is a technique that rescales the data to a common range, such as 0 to 10, to make it comparable across different units or scales. By applying these two steps, we obtained risk scores that are more robust and consistent for this dimension.

2.17 Deforestation (environmental)

Risk definition: Deforestation refers to the decrease in forest areas that are cleared for other uses such as agriculture, urbanization, or mining activities. 

Data source: Global Forest Resources Assessment 2020 (link to the dataset). The report considers forest losses and gains. A total of 236 countries and territories are included. The data points used are the net annual change from 2010 – 2020 in % and area.

Data attributes:

• Data level: Regional
• Data scope: Generic (not product specific)
• Source update frequency: Annually

Description of the data: This comprehensive report from FAO covers various aspects of the world’s forests. The Global Forest Resources Assessment (FRA) 2020 is the result of a collective effort by the FAO Forestry Department, FAO member countries, and institutional and resource partners.

Notes on the dataset: Any positive values (gains) are assigned a 0-risk score. The data for this dimension was transformed into risk scores from 0 to 10 using two steps: windsorization and normalization. Windsorization is a technique that replaces extreme values with more moderate ones, to reduce the impact of outliers on the analysis. Normalization is a technique that rescales the data to a common range, such as 0 to 10, to make it comparable across different units or scales. By applying these two steps, we obtained risk scores that are more robust and consistent for this dimension.

2.18 Soil degradation (environmental)

Risk definition: Soil degradation is the physical, chemical, and biological decline in soil quality. Causes of soil degradation include agricultural, industrial, and commercial pollution; loss of arable land due to urban expansion, overgrazing, unsustainable agricultural practices; and long-term climatic changes. 

Data source: The European Soil Data Centre (link to the dataset) provides modeled data on soil degradation at a regional level.  

Data attributes:

• Data level: Regional
• Data scope: Generic (not product specific)
• Source update frequency: 2012 is the latest version. 

Description of the data: This primary data can be used to calculate a risk score. The data for each pixel is the output of Version 1.1 of the JRC/ UniBasel "RUSLE-based Global Soil Erosion Modelling platform (GloSEM)". Although soil degradation includes more than just soil erosion, it is suggested that soil erosion is a good proxy of levels of overall soil degradation. 

Notes on the dataset: The data for this dimension was transformed into risk scores from 0 to 10 using two steps: windsorization and normalization. Windsorization is a technique that replaces extreme values with more moderate ones, to reduce the impact of outliers on the analysis. Normalization is a technique that rescales the data to a common range, such as 0 to 10, to make it comparable across different units or scales. By applying these two steps, we obtained risk scores that are more robust and consistent for this dimension.

2.19 Overall water risk (environmental)

Risk definition: The Overall Water Risk considers the physical quantity, physical quality, regulatory, and repetitional water-related risks of a region. 

Data source: The WWF Water Risk Filter includes a broad range of physical quantity, physical quality, regulatory, and repetitional water risk indicators.  The downloadable data is not currently available, as the tool was relaunched.

Data attributes:

Data level: Regional

Data scope: Generic (not product specific)

Source update frequency: Annually 

Description of the data: The Water Risk Filter covers various risk areas including physical risk, regulatory risk, and reputational risks relating to water. The underlying primary data is a combination of model simulation results, predictions, assessments, and satellite data. An overall basin risk score is calculated by aggregating the physical, regulatory, and reputational risks for a basin. In turn, these sub-categories are calculated by consolidating various index scores. For example, water scarcity (a physical risk) is calculated by consolidating scores for seven reputable indices (Aridity Index, Drought Frequency Probability, etc.). 

Notes on the dataset: The data for this dimension was transformed into risk scores from 0 to 10 using two steps: windsorization and normalization. Windsorization is a technique that replaces extreme values with more moderate ones, to reduce the impact of outliers on the analysis. Normalization is a technique that rescales the data to a common range, such as 0 to 10, to make it comparable across different units or scales. By applying these two steps, we obtained risk scores that are more robust and consistent for this dimension.

2.20 Ozone-depleting substances (environmental)

Risk definition: Ozone-depleting substances (ODS) are manufactured gases that destroy ozone once they reach the ozone layer. The ozone layer sits in the upper atmosphere and reduces the amount of harmful ultraviolet radiation that reaches Earth from the sun. Ultraviolet radiation can have detrimental effects on both humans and the environment, for instance, inducing skin cancer and cataracts, distorting plant growth, and damaging the marine environment.

Data source: United Nations Environment Programme Consumption of Controlled Substances

Description of the data: ODS consumption is reported in Ozone Depleting Potential (ODP) tonnes. There are individual datasets for each of the annexes: A/I, A/II, B/I, B/II, B/III, C/I, C/II, C/III, E/I, F. Negative values for a given year imply that quantities destroyed or quantities exported for the year exceeded the sum of production and imports, implying that the destroyed, or exported quantities came from stockpiles. Members of the European Union do not report their consumption; that is reported in aggregated form for all the member states.

Data attributes: 

• Data level: Regional
• Data scope: Generic (not product specific)
• Source update frequency: Annually 

2.21 Institutional arrangements / Good governance (governance)

Risk definition: Governance consists of the traditions and institutions by which authority in a country is exercised. This includes the process by which governments are selected, monitored, and replaced; the capacity of the government to effectively formulate and implement sound policies; and the respect of citizens and the state for the institutions that govern economic and social interactions among them. While governance may be seen in narrow political and administrative terms as decision-making by “the government,” good governance requires that all institutional actors involved in managing the social and environmental performance in a country, including citizens, organisations, and private entities, work in a common direction. Poor governance leads to increased political and social risks, institutional failure, and lowered capacities to deliver. Therefore, good social and environmental governance requires clear legal frameworks, comprehensive social and environmental policies, enforceable regulations, institutions that work, smooth execution, and citizen-based mechanisms of accountability, as well as strong interconnections between these entities. 

Recommended data source: The Worldwide Governance Indicators (WGI) database (link to the dataset) Ranks a variety of good governance indicators: government effectiveness; voice and accountability; political stability and absence of violence/terrorism; regulatory quality; rule of law; and control of corruption.

Data attributes:

• Data level: Country
• Data scope: Generic (not product specific)
• Source update frequency: Annually 

Description of the data: The WGI project provides aggregate and individual governance indicators for over 200 countries and territories over the period 1996–2020. The WGI gives a ranking per country with a percentile rank (0-100), and it can compare the rank of a country among all countries in the world. 0 corresponds to the lowest rank, and 100 corresponds to the highest rank. The ranking is based on a variety of credible sources. On average, at least 7 or more data sources are used per country. 

Notes on the dataset: The data for this dimension was transformed into risk scores from 0 to 10 using two steps: windsorization and normalization. Windsorization is a technique that replaces extreme values with more moderate ones, to reduce the impact of outliers on the analysis. Normalization is a technique that rescales the data to a common range, such as 0 to 10, to make it comparable across different units or scales. By applying these two steps, we obtained risk scores that are more robust and consistent for this dimension.