¹: World Bank, World Development Indicators data portal.

²: UNESCO, UIS statistics.

³: World Economic Forum, 2020b.

⁴: UNFPA, State of World Population 2020.

⁵: Ibid.

⁶: These gaps are computed as 1-(women share/men share) and represent the gender gap yet to close on this aspect in these countries.

⁷: All global and regional aggregates are computed as population-weighted averages. The aggregate scores reported here refer to constant sample population-weighted averages, which are based on the performance of the 107 countries covered continuously between 2006 and 2021.

⁸: With respect to parental leave, the OECD’s Employment database reports 2018 data. Paternity leave regulation has been updated in Switzerland via the Swiss “Votation Populaire” of the 27 September 2020 (source: www.admin.ch/gov/fr/start/dokumentation/abstimmungen/20200927.html).

⁹: ILO, 2020.

¹⁰: Pissarides 1992.

¹¹: Davis and von Wachter, 2011.

¹²: PwC, 2016.

¹³: ILO, 2021. Data from the International Labour Organization is consistently able to cover the first three quarters of 2020.

¹⁴: UNDP and UNWomen, 2020.

¹⁵: Alon, et al, 2020a.

¹⁶: Mooi-Reci and Risman, 2021.

¹⁷: Dingel and Neiman, 2020.

¹⁸: Data from the LinkedIn Economic Graph provides an overview of the entirety of 2019 and 2020 for a subset of economies.

¹⁹: World Economic Forum, 2020.

²⁰: UNICEF, 2020.

²¹: Sevilla and Smith, 2020.

²²: Collins, et al, 2021.

²³: Landivar, et al, 2020.

²⁴: Quarterly labour force participation gender parity is derived from ILO, ILOstat database; school closure and work closure indexes are obtained from the Oxford’s Coronavirus Government Response Tracker, Childcare public expenditure to GDP is obtained from the OECD social expenditure database and the flexibility of working arrangements is obtained from the World Economic Forum’s Executive opinion survey.

²⁵: Pre-existing childcare provision proxied by public expenditure on childcare.

²⁶: Konkel, 2021.

²⁷: World Economic Forum, 2020.

²⁸: To understand the skill profile of each occupation, analysts first identified a list of the most representative skills associated with an occupation, based on LinkedIn’s Skills Genome Metric, which calculates the ‘most representative’ skills across roles, using the TF-IDF method. The aggregate skills similarity between two occupations is then calculated as the cosine similarity between the skills profiles of those two occupations.

²⁹: Misa, 2011, Cardador, 2017, and Bian, et al, 2017.

³⁰: Fernandez-Mateo and Kaplan, 2018, and Ensmenger, 2010.

³¹: Averkamp, et al, 2020.

³²: Chernoff and Warman, 2020.

³³: UNDP and UNWomen, 2020.

³⁴: Data cited is taken from the World Bank’s Women, Business and the Law Database, 2021. Maternal and parental leave as well as additional reserve for men and women have been aggregated to calculate maximum available leave. We then estimate the theoretical ratio of days available to men and women. Those figures are compared with the full-time pay equivalence to account for the cost of taking care leave on families.

³⁵: Addati, et al, 2018, and ITUC, 2016.

³⁶: World Economic Forum, 2020.

³⁷: For some further discussion of this point see Hausmann, 2016.

³⁸: Following a methodology originally developed by the United Nations Development Programme (UNDP), the Global Gender Gap Index estimates the average income earned by women, relative to income earned by men, in a calculation that takes into account a country’s GDP per capita (US$), the share of women and men in the labour force, and their mean nominal wages. To account for globally rising income levels, beginning with 2018’s edition, the report no longer caps the maximum income per capita value considered in the calculation. This follows UNDP’s own adjustment of the methodology and the fact that the US$40,000 cap formerly used in previous editions of the Global Gender Gap Index had increasingly lost some of its ability to discern the level of gender-based income disparities among high-income nations such as the Nordics, the United States and the member states of the Gulf Cooperation Council. For a full overview of the 2016 methodology change, please refer to that report edition’s Appendix D.

³⁹: For more information about the Executive Opinion Survey, see World Economic Forum, 2020, Appendix B.

⁴⁰: Beginning with last year’s edition, the report utilizes the United Nations Population Division’s World Population Prospects as its sole reference source for the sex ratio at birth indicator. Previous editions of the report had utilized data from the U.S. Central Intelligence Agency’s World Factbook as an alternative data source. With the exception of Haiti, Kazakhstan and Pakistan, discrepancies in reported values between the two data repositories are no more than 2% for all countries covered by the index, resulting in minimal score differences in all cases.

⁴¹: This ratio is based on what is considered to be a “normal” sex ratio at birth: 1.06 males for every female born. See Klasen and Wink, 2003.

⁴²: This ratio is based on the standards used in the UNDP’s Gender-Related Development Index, which uses 87.5 years as the maximum age for women and 82.5 years as the maximum age for men.

⁴³: A first attempt to calculate the gender gap was made by the World Economic Forum in 2005; see Lopez-Claros and Zahidi, 2005. The 2005 index, which was attempting to capture women’s empowerment, used a “feminist” scale that rewarded women’s supremacy over men (highest score is assigned to the country with the biggest gap in favour of women).

⁴⁴: As in previous editions of the index, weights derived for the 2006 index were used again this year to allow for comparisons over time. They may be revised in future editions to reflect the evolution of the gender gap over the past decade.

⁴⁵: This is not strictly accurate in the case of the Health and Survival subindex, where the highest possible value a country can achieve is 0.9796. However, for purposes of simplicity, we will refer to this value as 1 throughout the chapters and in all tables, figures and Country Profiles.

⁴⁶: Because of the special equality benchmark value of 0.9796 for the Health and Survival subindex, it is not strictly accurate that the equality benchmark for the overall index score is 1. This value is in fact (1 + 1 + 1 + 0.9796) / 4 = 0.9949. However, for purposes of simplicity, we will refer to the overall equality benchmark as 1 throughout the chapters and in all tables, figures and Country Profiles.

⁴⁷: Since the indicators in the subindexes are weighted by the standard deviations, the final scores for the subindexes and the overall index are not a pure measure of the gap vis-à-vis the equality benchmark, and therefore cannot be strictly interpreted as percentage values measuring the closure of the gender gap. However, for ease of interpretation and intuitive appeal, we will be using the percentage concept as a rough interpretation of the final scores.

⁴⁸: International Standard Classification of Occupations (ISCO), www.ilo.org/public/english/bureau/stat/isco/.

⁴⁹: International Standard Classification of Education (ISCED), www.uis.unesco.org/Education/Documents/isced-2011-en.pdf.

⁵⁰: Please note that the parity benchmark is 1 for all indicators, except sex ratio at birth (0.944) and healthy life expectancy (1.06). As a consequence, gender parity for the Health and Survival subindex corresponds to a score of 1.002 rather than 1.