Enabling Women's Economic Empowerment: An AI Driven Approach to Gender Equality

At a glance




2022 – 2023


Mexico City’s Government, Women’s Ministry, BMZ, GIZ, SEMUJERES CDMX, White Ribbon Alliance, Equidad Organizacion Feminista, ProsperIA, Ethos Innovación en Políticas Públicas


gender equality, women’s financial independence, unpaid care work, machine learning, natural language processing


Survey and Census Data, Crowdsourced Data (processed through a natural-language understanding (NLU) Model) and Satellite Imagery (when lack of up-to-date data available)


SDG #5

The Challenge

Globally, women perform 76.2 per cent of total hours of unpaid care work - more than three times as much as men. This has direct implications on their opportunities to participate in other social, cultural and economic activities. Men in Mexico dedicate the second least time in unpaid care work globally, behind only Chile. COVID-19 meant that women took on additional domestic work, further constraining their opportunities to engage in gainful employment. While evidence suggests that a better care ecosystem can advance women’s economic participation, policymakers often lack concrete information, for example, on infrastructure demands, that can help build such an ecosystem.

The Approach

Mexico's City Government, in collaboration with GIZ, is developing an intelligent platform that integrates various data sources, including Natural Language Understanding (NLU) processed data, with administrative records related to the care system, information on economic opportunities and enablers of women’s participation such as transportation and childcare. Further data types can be added in future. This platform will enable decision-makers to make informed and data-driven decisions quickly and easily.

The Benefits

Cost-efficiency: The platform utilizes crowdsourced data, which can be an inexpensive way to gather information.

Innovation: The use of an innovative NLU model to process information allows for the inclusion of open-ended questions that provide complementary information to traditional surveys.

Flexibility: The platform can incorporate as many data sources as needed or are available, providing flexibility in terms of data analysis.

Scalabilityand Adaptability: The NLU model can be used in all Spanish-speaking countries and contexts.



of total unpaid care work is performed by women globally.

The Context

Mexico City's Data Gap: The government collects data systematically but due to the city’s size and diversity, not all the needs of the community are fully captured, particularly those of women.

The government wants to understand women's desires, expectations and limitations in relation to paid work, both formal and informal, at a local level to improve its programs and policies. This is especially important in the context of the disproportionate burden of unpaid care and domestic work that women face.

Accessible, affordable and high-quality domestic and care services can help move these tasks into the paid sector and increase participation in social, political, and economic spheres of both those who perform the tasks and those – mostly women – whose time is then freed up. Lack of such services can limit women’s choices and opportunities. In urbanized areas like Mexico, where nearly 40% of women have children, access to childcare is vital to facilitate their participation in social and economic spheres.

How can better data contribute to better policy?

The Women’s Ministry of Mexico City, with support from Data4Policy (a policy initiative of the German Federal Ministry for Economic Cooperation and Development (BMZ)), designed a prototype that uses different types of data, such as census, survey, administrative records and crowdsourced data to support policymaking.

This project overall comprises three primary components.  

1) Conducting localized analysis using various data sources.  

2) The extensive collection and processing of large amounts of data.  

3) A comprehensive analysis of all gathered and processed data, which aims to generate actionable policy recommendations.  

Such recommendations will also consider the automatically generated recommendations created by the spatial-intelligence algorithm integrated in the platform.

Figure 1 shows how the platform integrates different data sources.  

Figure 1. Integration of different methods and data sources in the platform

Source: ProsperIA 2023

Figure 2 displays the platform dashboard, illustrating the spatial distribution of answers as categorized by the NLU.

Figure 2. Most recurring topic among all responses within a ZIP code.

Source: ProsperIA IncluIA – GIZ SEMUJERES Map. 2023.

Women and Care Work

As an example, if the government would like to know where women who have indicated the need for a caregiver are distributed geographically, this information can be filtered as shown in Figure 3.

Figure 3. Spatial distribution of the answers where ”Having a person under my care” was ranked in the top 6 factors influencing the person’s answer to the question ”What do you most want or need to find employment or better employment?”

Source: ProsperIA IncluIA – GIZ SEMUJERES Map. 2023.

Attention to this issue can now be targeted to specific zip code areas with lower incomes and a heightened proportion of economically inactive women. This can be accomplished by employing the filters depicted in Figure 4.

This shows that the shaded areas primarily coincide with outlying areas far from the economic hub of Mexico City. Noteworthy regions include the Magdalena Contreras, Milpa Alta, and Iztapalapa municipalities.

Figure 4. Spatial distribution of the answers where “Having a person under my care” was ranked in the top 6 factors influencing the person’s answer to the question ”What do you most want or need to find employment or better employment?” (filtered for income below 22,000 pesos per quarter average and percentage of economically inactive women between 30 and 60 percent)

Source: ProsperIA IncluIA – GIZ SEMUJERES Map. 2023.

Policymakers then gains the option to incorporate pertinent supply-of-care and labor-supply factors as additional criteria. These may include considerations like proximity to the nearest subway station or public childcare centers. Figure 5 shows localities listed in figure 4 where the distance to the nearest childcare center falls between 2.5 and 4.7 km.  

Figure 5. Spatial distribution of the answers where “Having a person under my care” was ranked in the top 6 factors influencing the person’s answer to the question ”What do you most want or need to find employment or better employment?” (filtered for income below 22,000 pesos per quarter average and percentage of economically inactive women between 30 and 60 percent and distance to nearest available public child care facilities and metro stations)

Source: ProsperIA IncluIA – GIZ SEMUJERES Map. 2023.

In the lower section of the left panel, the policymaker can also opt to reveal automated suggestions from a spatial intelligence algorithm, pinpointing the optimal locations for any potential new childcare centers (see Figure 6).

Figure 6. Recommended new childcare facilities

Source: ProsperIA IncluIA – GIZ SEMUJERES Map. 2023.

The platform presents four potential locations for new childcare centers. To determine which center to prioritize, factors such as child coverage and the presence of economically inactive women come into play. In this instance, by employing the filters in the right panel, we can opt to view recommendations that optimize for these factors.

At the site of the top recommendation, located in the Iztapalapa neighborhood, an estimated 3029 children could benefit from a new childcare center. In this locality, nearly 50% of women are economically inactive, and there are 4264 households headed by females.

Source: ProsperIA IncluIA – GIZ SEMUJERES Map. 2023.

The platform's recommendations, tailored to prioritize child coverage and the needs of economically inactive women to access gainful employment, exemplify its potential to promote more fair policy outcomes.  

Childcare facilities are just one example of how the prototype can be scaled according to the crowdsourced data on women’s needs. Information layers regarding elderly care, transportation routes (not only metro lines) and other critical factors can be included depending on the information retrieved from the crowdsourced data.  

Where do we go from here?

In the third stage of the prototype, there are three main goals:

  1. Policy Analysis. The key insights and findings are being processed and consolidated in a policy analysis report that will be available in an interactive form on the open-source dashboard. As part of this component, there will be capacity building activities to teach policymakers how to draw insights from data and translate them into policy.  
  1. Data insights for policy making. By the end of the year, we plan to host an event with representatives from various government entities. The key findings will be presented and we will demonstrate how ministries can benefit from incorporating this information into their budgeting processes for the 2024 fiscal year.
  1. Capacity Building. In the second half of the year, we will hold in-person capacity building events where the use case prototype and data will be handed over to technical personnel from Mexico City's government.  

Case Downloads

New ways of using data for gender equality
by BMZ
explore more Use Cases