Missing SDG Gender Indicators

The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.


Introduction
The Sustainable Development Goals (SDGs) lay out an ambitious agenda including that of achieving gender equality by 2030.This agenda is paired with a set of goals and targets measured by concrete indicators and is adopted by nearly all countries.SDG 5 focuses on gender equality and sets 9 measurable targets (with 14 indicators) on issues that especially affect women and girls (United Nations, 2022).But gender cuts across a far wider range of the SDGs than just the indicators under Goal 5.For example, SDG 3 on ensuring good health and well-being includes a target on reducing maternal mortality (target 3.1).The SDG agenda also calls for sex disaggregated data across several goals where monitoring of gender disparities is essential for effective policy.
For example, SDG 8 on promoting decent work and economic growth sets a target of achieving full employment and equal pay for all women and men (target 8.5).These gender data are promoted as key to understanding if and how patterns of progress differ between women and men or girls and boys (UN Women, 2022).Countries are, however, falling short on reporting on gender-related indicators of the SDGs.This paper analyzes the patterns underlying these data gaps.
The objective of this paper is to look at this globally agreed-upon set of gender data indicators and identify key country patterns related to the existence, or lack of, such data.We focus on the availability of data reporting on the SDGs since they represent an internationally agreedupon set of goals to meet and for countries to report on (UNSD, 2022).
Missing gender data is not a new concern.There are different approaches to diagnosing the causes of the lack of gender data.One approach put forth by Bonfert et al (2022) emphasizes four obstacles to more gender data (Figure 1): (i) lack of data sources such as core and/or specialized surveys, censuses or relevant administrative data (that is, the data simply are not collected); (ii) methodological flaws in data collection (e.g., collecting land holdings of households but not identifying which household member has the rights/ownership to this land); (iii) insufficient processing of existing data; and (iv) lack of dissemination even when data are available and processed.
Figure 1: Sources of gender data gaps Source: Bonfert et al. 2022 Related, but not identical, Buvinic Furst-Nichols, and Koolwal (2014) discuss gender data gaps as driven by four gaps: (i) lack of regular production at the country level; (ii) lack of international standards; (iii) lack of information across domains; (iv) lack of granularity, i.e., lack of large, detailed datasets making possible disaggregation.
In this paper we look at the production and reporting of SDG indicators on gender, clearly laying out the availability (alternatively the lack of) data along indicator typesuniquely gender focused versus cross-cutting.We then focus on the challenges posed by insufficient processing of available data or the lack of dissemination even when processed data and constructed indicators are available.While focusing on improving the statistical systems is an important part of the agenda to fulfill the goal of reporting on gender-related SDGs, some rapid improvements can be made from existing data.

Gender indicators for the SDGs
Although nearly all countries have agreed to report on the SDG indicators, major gaps exist in indicator availability since the SDG agenda's inception in 2015 (Dang and Serajuddin 2020 (32 of the 50), the 18 others are related to goals specific to females-highlighting that genderrelated SDG indicators are not only about sex disaggregation.
Table 1 shows the share of countries for which there is at least one annual data point in the five-year period from 2016 to 2020 for each of the 50 gender-related indicators.For the 14 indicators under Goal 5, the country average availability is 37%, only marginally higher availability compared with the average availability for all gender-related indicators.Figure 2 shows this distribution.No country has more than 10 of these 14 indicators in the 5-year period.
Forty-one countries report three or fewer indicators.Annex 1 presents the availability of SDG5 gender-related indicators by region and country income grouping.

Figure 2. Availability of SDG5 gender-related indicators (N=181 countries)
Note: The figure shows the coverage of the 14 SDG 5 indicators where coverage is defined as having at least one annual data point in the five-year period from 2016 to 2020 for an indicator as compiled by the UN.
Among indicators that require sex disaggregation (32 out of the 50), both the population data and the sex-disaggregated data are not reported for any country for five indicators (such as for indicator 10.2.1).For four of these 32 indicators, the sex-disaggregated and population coverage rates match, as we would expect if the underlying data identified individual sex, was collected for both males and females, and was processed accordingly.We would not expect the sex-disaggregated coverage rate to exceed the population coverage, and it never does.Moreover, if the country has a sex-disaggregated data point they also have a population estimate.But the reverse is not the case.For six of these 32 indicators (19%), while there is some reported data for the population indicator, there is no sex-disaggregated data reported.As an example, 56% of countries report a population rate for SDG indicator 10.2.1 (the proportion of people living below 50 percent of median income), yet no country reports this statistic by sex.These missing data are not the result of missing sex in underlying data source (in this case, household surveys).The measure itself (living below an income threshold) is defined at the household level and so one can produce a sex-disaggregated estimate based on the households in which individuals reside.For example, Munoz Boudet et al. (2021) report poverty rates by sex.6 In the remaining 17 cases (out of 32), where there is some reported data for both the population and by sex-disaggregation, in a handful of cases there are large gaps between the percentage of countries with a recent value by sex and those reporting a population estimate (i.e. comparing the last two columns in Table 1 when both columns are non-zero).These results show that the problem of missing SDG gender measures is, in part, a problem with processing existing data rather than the lack of the primary data collection, since the underlying sources (typically household surveys) almost always (if not always) collect the sex of household members.If countries with a population estimate also reported data by sex, the SDG gender coverage rate would rise from 31% to 43%.
Evidence from other sources underscore the problem of available data not being reported.
In a review of national statistics for 12 countries related to sex-disaggregated data on asset ownership, employment, and entrepreneurship, Bonfert et al (2023)  Notably, the high-income countries do not have higher coverage of gender-related SDG indicators (Figure 4).GDP per capita is not associated with better coverage of gender statistics in the UN SDG database (Figure 5).This is also the case for SDG indicators overall: high income countries do not have higher rates of reporting of all 181 SDG indicators (rates are high income 64%, upper middle income 72%, lower middle income 70%, and low income 65%).Yet, high income countries perform notably better on the Statistical Performance Indicators and Index (SPI) -the World Bank's new official tool to measure country statistical capacity and is being added to the SDG indicators under SDG17 (Dang et al. 2023).
One explanation for this paradox on reporting SDGs and statistical strength overall is that    Note: Percent of countries with any reporting on the indicator in the five years (2016)(2017)(2018)(2019)(2020).NA indicates that the indicator is not relevant in regards to sex disaggregation.Source: UN SDG Global Database.https://unstats.un.org/sdgs/dataportal

SDG gender indicator availability and country statistical performance
Next we assess how a country's overall statistical performance relates to the availability of gender data, and identify countries that may have strong systems overall but are underperforming on gender statistics.To do so, we compare the availability of gender statistics to scores on the World Bank's Statistical Performance Indicators (SPI) (Dang et al 2023).
The World Bank's Statistical Performance Indicators (SPI) measure statistical performance for 174 countries covering over 99% of the world population.The indicators are grouped into five pillars: (1) data use, which captures the demand side of the statistical system; (2) data services, which looks at the interaction between data supply and demand such as the openness of data and quality of data releases; (3) data products, which reviews whether countries report on global indicators;8 (4) data sources, which assesses whether censuses, surveys, and other data sources are created; and (5) data infrastructure, which captures whether foundations such as financing, skills, and governance needed for a strong statistical system are in place.Within each pillar is a set of dimensions, and under each dimension is a set of indicators to measure performance.The indicators provide a time series extending at least from 2016 to 2020 in all cases, with some indicators going back to 2004.9The indicators are summarized as an index, termed the SPI overall score, with scores ranging from a low of 0 to a high of 100.We use the SPI data for 2019.
There is a positive relationship between countries' SPI overall scores and the availability of SDG gender indicators (Figure 6).For pillar 3, which is overall SDG coverage, likewise there is a positive relationship with SDG gender indicators (Figure 7).This is not surprising since pillar 3

Discussion
The For 23 indicators, a lack of data reporting seems to be a cause of missing SDG gender indicators.
In these instances, population estimates are being reported, but the sex-disaggregated counterpart is not reported to the same degree though this disaggregation, if not always, is nearly always feasible.This gap in SDG gender indicator reporting seems to be low-hanging fruit, addressed by ensuring that sex-disaggregated information is processed and reported.
For the other indicators, we cannot as easily disentangle if data reporting is the source of the problem or, rather, the lack of relevant surveys/administrative data.For seven indicators, there are no countries with any data point in the five-year period.Statistical system strength as measured by the SPI score is positively correlated with SDG gender data availability; better statistical systems are an important part of the solution.Still, when assessing the performance of gender data availability by a country income level, poor countries are not doing worse despite often weaker statistical systems.The SDG agenda was set in a way that all countries, irrespective of their development status or income level, were to report their progress on all targets, which is in contrast with the MDGs era, which was largely focused on low-and middle-income countries.Highincome countries may have been slow to adjust to this shift in the agenda and have, therefore, underreported statistics (MacFeely, 2018).Certainly there may also be cases where they do not collect certain statistics because of the lack of relevance to their country contexts, as noted in the case of child marriage and female genital mutilation, but we do not find evidence to support this as a driving factor of the result that poorer countries do as well as high-income countries in reporting SDG gender-related indicators.
Beyond these factors, we are left with unexplained variation in gender data availability across countries.This is partly captured in the notable over (and under) performance in reporting gender-related SDGs relative to statistical system strength.One can take a somewhat optimistic perspective in combining this with the two previous findingsthat some portion of underreporting is not driven by lack of data but by under-reporting, and that country income is not driving higher rates of reporting.Even without major investments in statistical systems or the years it may take for such investments to yield results, with some concerted effort, it is possible to achieve big wins in SDG gender indicator coverage.
Meanwhile, it is important to note that while the SDG framework offers the world a consensus set of indicators selected as part of a global consultative process, there are other important country-level gender-related indicators available outside the SDG system.Sources such as the UN Women Data Hub and the World Bank's Gender Data Portal offer compilations of national statistics produced by countries and curated by international agencies.
For SDG indicator 1.3.1, on the proportion of population covered by social protection floors/systems, 79% of countries have a recent value for the population, but only 8% of countries have a sex disaggregated data point.A less drastic example is SDG indicator 4.1.1,related to early childhood education: 64 percent of countries have a population estimate for this indicator but only 53 percent have an estimate by sex.
richer countries may have been slow to report SDGs compared with low-and middle-income countries that have experience in engaging with the Millennium Development Goals (MDGs) (MacFeely 2018).A second explanation lies in the specific focus of some SDGs.This difference in the overall performance of national statistical systems and the reporting on gender SDGs might be explained by presence in the latter of indicators which relate to phenomena that are arguably infrequent or rare for high income countries (or perceived as such).For example, data on child marriage and on female genital mutilation (covered in SDG target 5.3) are rarely collected in OECD countries.OECD (2022)  describes the extra lengths needed to get such data from alternate sources in order to be able to report on this SDG.A third explanation, related to the second one above, is the presence of systematic and large-scale data collection under the Demographic and Health Survey (DHS) and the Multiple Indicator Cluster Survey (MICS) programs, which are focused on low-income countries (and often financed with non-national resources).These surveys are often the source of gender-related data, especially in the domains of female health and empowerment.To assess this, we examine the main data sources for the 50 SDG gender indicators.The DHS or MICs is the source for at least one country data point for 13 out of 50 indicators but only extensively (well over half of the data points) for 5 indicators.7We find very slight evidence that the DHS/MICs data source explains lower coverage of gender-related SDGs in high-income countries relative to lower income countries.When excluding these 5 DHS/MICs-dominant indicators, high-income countries have basically the same coverage (31%) as low (22%) and lower middle (31%) countries.And they continue to lag behind upper middle income countries (35.4%).

Figure 3 .
Figure 3. Availability of SDG gender indicators by region (N=181 countries)

Figure 4 .
Figure 4. Availability of SDG gender indicators by income group (N=181 countries)

Figure 8 :Figure 9 :
Figure 8: Top 15 over/under performers on availability of gender SDG indicators in terms of number of indicators

Figure
Figure A1.3.Availability of Tier 2 SDG gender indicators by region (N=181 countries) ). Gender-related SDG indicators are no exception.There are 231 unique SDG indicators.Many of the indicators, even if not obviously related to gender, nonetheless have sub-indicators, such as, indicators by sex, age, or disability status.The UN global SDG indicators database provides accessto the data compiled for tracking progress toward fulfilling the SDGs.We use this data source, not in the UN Women list.Relatedly, Open Data Watch (2019) refers to 32 SDG gender indicators and another 36 "additional" SDG gender indicators.The difference between their 68 and our 50 SDG gender indicators is that some of theirs are, in our assessment, gender neutral in terms of the present drafting of the indicator (such as 1.5.1 Number of deaths, missing persons and directly affected persons attributed to disasters per 100,000 population). 2Tier 1 indicators, according to the UN, are indicators that are conceptually clear, have an internationally established methodology and standards available, and data are regularly produced by countries for at least 50% of countries and of the population in every region where relevant.Tier 2 indicators are conceptually clear and have an internationally rather than individual NSO websites, because data submitted to the UN Global SDG monitoring database goes through a standardized process including a certain level of quality control and documentation review.We explore the coverage of the 50 gender-related SDG indicators out of the 231 unique indicators. 1 As noted earlier, gender-related SDG indicators are not limited to SDG 5 on gender equality, but rather span indicators across 10 out of 17 of the SDG goals.All 50 SDG-gender indicators are Tier 1 or 2 SDG indicators. 2 While most are related to sex disaggregation of data 1 These 50 indicators closely match the UN Women minimum set of 52 quantitative gender indicators from the SDGs (United Nations Economic and Social Council 2012), subsequently revised to be 51 quantitative indicators, with a few exceptions.These exceptions are: (i) indicators 4.7.1, 4.a.1, and 13.3.1 are in the UN Women minimum set but not here as in our view they are not gender-related or sex-disaggregated measures.(ii)indicator 1.1.1includesthe "working poor" (employed population below international poverty line) by sex component which is inTable 1 but established methodology and set of standards, but are not regularly produced by countries.There are no gender equality indicators in the third category, Tier 3, which is defined as an indicator with no internationally established methodology or standards established, and, thus, these indicators are likely to have the lowest rate of coverage.(UNSD, 2022).

Table 1 . SDG indicators related to gender (N=181 countries) Goal Indicator Tier Description Sub-Indicator (if any)
the SDG gender indicators itself.Interestingly, although SPI and the SDG gender indicator coverage are positively correlated and SPI is positively correlated with country income level (not shown), as noted earlier, SDG gender indicator coverage is slightly negatively correlated with country income.A breakdown of correlations is reported in Annex 2.It is among the countries in the poorest quintile of statistical system scoring where the gap in gender data availability is largest.Among the countries within 2 nd , 3 rd , 4 th , or top quintile of the SPI score, the mean gender-related SDG indicators availability is between 17 and 20.However, for the countries in the bottom quintile of the SPI score, only 12 gender indicators out of 50 are available on average.

Table 2 . Regression of availability of SDG gender indicators on country traits
Table2reports on regressions controlling for some additional country-level traits.These include three measures of gender equality: World Bank Women, Business and the Controlling for region, the correlations between SDG gender indicators and both SPI and GDP per capita still hold.There is some indication that countries with worse gender inequality measures fare better in terms of SDG gender indicator reporting.: *** indicates statistical significance at 1%, * at 5% and + at 10%.Constant term included.The WBL has a range of 1-100, with a high score indicating more gender equal laws and regulations.The OECD SIGI SIGI measures the extent of gender discriminatory legislation and restrictive social norms and practices, where a high score indicates greater gender inequality.Likewise, the GII measures poor outcomes for women in regards to reproductive health, empowerment, and the labor market, and a high score indicates greater gender inequality. Notes world's Sustainable Development Goal (SDG) agenda lays out an ambitious set of indicators to track progress.While the overall coverage of the 231 indicators certainly needs improvement, the coverage for the 50 gender-related indicators is especially low.On average, countries have 30% of these indicators for at least one year between 2016 and 2020, compared to a rate of 65% for all 181 SDG indicators.Moreover, the subset of 14 indicators under the specific Goal 5 on gender equality fare only slightly better.This low coverage is not a problem of illdefined indictors.These 50 indicators are classified as either Tier 1 (18 indicators) or Tier 2 (32 indicators) in terms of statistical complexity; so the methodology to collect such data is established.Clearly, the world needs more reporting on gender-relevant indicators, but how much of this problem is one of lack of data (i.e.no survey exists) versus a failure in the reporting process?