CAF - Motherhood and Structural Change Insights from Rural Latin America

C A F - W O R KI N G PA P E R # 2 0 2 5 / 0 7

F i r s t v e r s i o n : A u g u s t 2 9 , 2 0 2 5

Women, Motherhood, and Structural Transformation. Insights from Rural Latin America Mariana Marchionni1 | Julián Pierino Pedrazzi2 | María Florencia Pinto3 1CEDLAS - IIE - Universidad Nacional de La Plata and CONICET. marchionni.mariana@gmail.com 2CEDLAS - IIE - Universidad Nacional de La Plata and CONICET. pedrazzi.julian@gmail.com 3CEDLAS - IIE - Universidad Nacional de La Plata. mflorenciapinto@gmail.com Structural transformation—the shift from agriculture to industry and services—is key to economic development, and can reshape labor market gender gaps. Yet little is known about how this process has unfolded in rural Latin America, where women face disadvantages from both gender and rurality. We document rural women’s labor market outcomes in 14 countries using harmonized household surveys, estimate motherhood effects using a pseudo-event study around first childbirth, and examine mechanisms using time-use data from Mexico. Despite educational gains, rural women still lag behind rural men and urban women in employment, hours, and earnings. While structural transformation has reduced informality and increased service and formal job participation, unpaid family work and precarious employment remain widespread among rural women. Motherhood further exacerbates disadvantages. Rural mothers face smaller employment drops than urban mothers, but are increasingly pushed into unpaid work and low-skilled self-employment. Evidence from Mexico shows this stems less from childcare than from heavier household chores, home production, and limited

access to labor-saving technologies. Our paper provides the first evidence on how structural transformation interacts with motherhood in rural Latin America, showing that structural change alone cannot ensure inclusive opportunities for rural women. K E Y W O R D S Structural transformation, child penalty, motherhood effect, gender inequality, Latin America We are very grateful to Hugo Ñopo, Dolores de la Mata, and Juan Odriozola for their valuable comments and suggestions. Findings, interpretations and conclusions expressed in this publication are the sole responsibility of its author(s) and cannot be, in any way, attributed to CAF, its Executive Directors or the countries they represent. CAF does not guarantee the accuracy of the data included in this publication and is not, in any way, responsible for any consequences resulting from its use. ©2025 Corporación Andina de FomentoC A F - D O C U M E NT O D E T R AB A J O # 2 0 2 5 / 0 7

E s t a v e r s i ó n : 2 9 d e a g o s t o d e 2 0 2 5

Mujeres, maternidad y transformación estructural. Evidencia en áreas rurales de América Latina Mariana Marchionni1 | Julián Pierino Pedrazzi2 | María Florencia Pinto3 1CEDLAS - IIE - Universidad Nacional de La Plata and CONICET. marchionni.mariana@gmail.com 2CEDLAS - IIE - Universidad Nacional de La Plata and CONICET. pedrazzi.julian@gmail.com 3CEDLAS - IIE - Universidad Nacional de La Plata. mflorenciapinto@gmail.com La transformación estructural—el paso de la agricultura a la industria y los servicios—es clave para el desarrollo económico

y puede reconfigurar las brechas de género en los mercados laborales. Sin embargo, se sabe poco sobre cómo este proceso se ha desarrollado en América Latina rural, donde las mujeres enfrentan desventajas tanto por género como por ruralidad. Este trabajo estudia su situación en 14 países de la región a lo largo del cambio estructural con encuestas de hogares armonizadas. Se muestra que aunque la transformación estructural redujo la informalidad y aumentó la participación en empleos formales y servicios, el trabajo familiar no remunerado y el empleo precario siguen siendo comunes entre las mujeres rurales. A partir de un pseudo-estudio de eventos, se muestra que el nacimiento del primer hijo agrava estas desventajas. La maternidad reduce menos el empleo de las mujeres rurales que el de las urbanas, aunque aumenta más su inserción en trabajos no remunerados y autoempleo de baja calificación. Evidencia de México sugiere que esto se debe al mayor peso de tareas domésticas, producción en el hogar y acceso limitado a tecnologías del hogar. Este trabajo es el primero en estudiar cómo la transformación estructural interactúa con la maternidad en la América Latina rural, mostrando que el cambio estructural por sí solo no asegura oportunidades inclusivas para las mujeres rurales. K E Y W O R D S Transformación estructural, penalidad por maternidad, efecto maternidad, desigualdad de género, América Latina Estamos muy agradecidos con Hugo Ñopo, Dolores de la Mata y Juan Odriozola por sus valiosos comentarios y sugerencias. Los resultados, interpretaciones y conclusiones expresados en esta publicación son de exclusiva

responsabilidad de su(s) autor(es), y de ninguna manera pueden ser atribuidos a CAF, a los miembros de su Directorio Ejecutivo o a los países que ellos representan. CAF no garantiza la exactitud de los datos incluidos en esta publicación y no se hace responsable en ningún aspecto de las consecuencias que resulten de su utilización. ©2025 Corporación Andina de FomentoMARCHIONNI ET AL . 2 1 | INTRODUCTION Structural change is a key feature of the development process, as economies transition from primary sector activities toward more diverse and industrialized sectors. This shift typically involves a decline in agricultural employment and an expansion of the manufacturing and service sectors (Gollin et al., 2002; Jonasson and Helfand, 2010; McMillan et al., 2014). In this context, labor organization also undergoes a transformation, moving from self-employment, particularly in agriculture, to salaried and formal jobs (Jensen, 2022; Schoar, 2010; La Porta and Shleifer, 2008, 2014; Faggio and Silva, 2014). This pattern is evident when comparing economic development across countries: as GDP per capita increases, self-employment declines, while salaried positions in both agricultural and non-agricultural sectors expand (Bandiera et al., 2022). Also, structural change can help close the gender gap in the labor market. As Dinkelman and Ngai (2022) explain, first, the expansion of the service sector can create more accessible employment opportunities for women. As economies shift away from agriculture and develop their service sectors, women may gain access to less physically demanding jobs that are more compatible with motherhood and potentially offer better wages. For instance, jobs in education, health, trade, and domestic services have the potential to absorb more

female workers, improving their labor conditions. Second, the marketization of domestic work through the growth of the service sector can provide substitutes for unpaid domestic labor. The availability of childcare services, house cleaning, and prepared meals can reduce the time women—specially mothers—spend on household tasks, thereby increasing their labor market participation. In developed countries and in the most developed areas of emerging economies— typically cities—similar trends have already facilitated greater inclusion in the labor market for women with children (Goldin, 1995; Blau and Kahn, 2017). However, this process is not automatic. For the emerging sectors to effectively absorb female workers, jobs must be accessible in terms of skills and working conditions, and key services that substitute domestic labor—such as childcare—must be both available and affordable. This paper explores how structural transformation in rural areas of Latin America is shaping women’s labor market decisions, employment structure, and the allocation of time between paid and unpaid work. By analyzing the evolution of women’s labor supply, the types of employment they engage in, and how they balance work and household responsibilities, we aim to understand how structural changes are affecting the gendered division of labor in rural Latin America and whether they contribute to the larger gender gaps observed in rural areas compared to urban areas. To do so, we rely on harmonized household surveys from the Socioeconomic Database for Latin America and the Caribbean (CEDLAS and The World Bank, 2023), which provide nationally representative samples disaggregated by urban and rural areas for 14 Latin American countries over the 2000-2023 period. Our analysis shows that rural women in Latin America face a persistent double disadvantage in labor markets, stemming from both their gender and their place of residence. Despite notable educational progress, their employment rate in 2023 was 56%, about eight percentage points lower than that of urban women and more than 30 points lower than

rural men. They also worked fewer hours—37 per week on average, compared to 41 for urban women and 45 for rural men—and earned significantly less—their monthly earnings were 21% below those of rural men and 27% below those of urban women. Importantly, structural transformation has contributed to rising participation in the service sector and in formal salaried jobs. Yet these shifts have not sufficed to close rural-urban and gender gaps: informal work arrangements continue to dominate women’s employment in ruralMARCHIONNI ET AL . 3 areas, especially in the form of unpaid family labor. In 2023, 68% of rural working women remained in informal employment, and about 16% were unpaid family workers—a share that rises to 30% within the primary sector. After providing this broader picture, we zoom in on the experiences of mothers by estimating the effect of children—motherhood effects or child penalties—on rural and urban women in Latin America using a pseudo-event study approach around the birth of the first child (Kleven, 2022; Kleven et al., 2024; Marchionni and Pedrazzi, 2025). We find that child penalties are large and persistent across the region, but their magnitude differs between rural and urban settings. Rural mothers experience smaller short-term drops in employment than urban mothers—18 percentage point drop compared to 22 points—, and show partial recovery in subsequent years. As a result, the rural-to-urban earnings ratio falls from 36% before motherhood to 27% afterward. However, this relative advantage comes at a cost: motherhood increasingly pushes rural women into unpaid work and lowskilled self-employment. For instance, motherhood increases the likelihood of unpaid work among rural women by about 10 percentage points, while in urban areas the increase is only 2 points, reinforcing structural disadvantages and widening income gaps with urban

women. Taken together, our findings underscore that structural transformation has not translated into inclusive gains for rural mothers, as caregiving constraints and occupational segregation still limit their ability to benefit from economic development. Complementary evidence from time-use data in Mexico highlights the domestic side of these constraints. While childcare demands are substantial—mothers of children aged 0-5 devote around 46 hours per week to childcare—the rural-urban gap in this dimension is small. Instead, rural mothers spend significantly more time than their urban peers on household chores, home production for their own consumption, and tasks such as water collection. These heavier domestic responsibilities, coupled with lower access to laborsaving technologies and external services, restrict the time rural mothers can devote to paid work. This evidence suggests that the persistence of rural–urban gaps in female employment is not only the result of labor market segmentation, but also of unequal time burdens within the household, which continue to weigh disproportionately on rural women. This paper contributes to at least two strands of literature. First, it builds on the literature linking structural transformation with women’s employment opportunities (e.g., Goldin, 1995; Dinkelman and Ngai, 2022). While this literature documents how the shift from agriculture to services has historically expanded female labor force participation, evidence for Latin America—and especially for rural areas of the region—remains virtually absent. Our paper fills this gap by providing a detailed account of how rural women’s labor market outcomes have evolved during the past two decades in 14 Latin American countries, using harmonized indicators that ensure comparability across time and space. Second, it relates to the growing literature on child penalties, which documents the persistent impact of motherhood on women’s labor market trajectories in developed and developing countries (Kleven et al., 2019a, 2024; Marchionni and Pedrazzi, 2025). Our contribution is to bring these strands of the literature together in the context of rural Latin America, a setting where

gender gaps are particularly wide and the structural transformation process is still ongoing. By combining harmonized cross-country data with a pseudo-event study approach, we provide the first systematic evidence on how structural change shapes both women’s labor market participation and motherhood effects in rural areas. In addition, by linking labor market outcomes to time-use data, we uncover how the unequal distribution of unpaid work within households constrains rural mothers’ ability to benefit from new economic opportunities. The remainder of the paper is organized as follows. Section 2 describes the data sources and explains the construction of the main variables. Section 3 provides a descriptiveMARCHIONNI ET AL . 4 overview of rural women’s position in Latin American labor markets, documenting their double disadvantage relative to rural men and urban women. Section 4 estimates motherhood effects in rural and urban areas using a pseudo-event study approach around the birth of the first child. Section 5 links these motherhood effects to the process of structural transformation, analyzing how sectoral shifts and changes in employment relations shape women’s opportunities. Section 6 goes beyond labor market outcomes to examine time-use patterns, childcare access, and domestic responsibilities, with a focus on Mexico, in order to shed light on the household mechanisms underlying rural women’s disadvantage. Finally, Section 7 discusses the main findings and concludes. 2 | DATA | Data sources and sample We rely primarily on the Socioeconomic Database for Latin America and the Caribbean (SEDLAC), a joint initiative of the Center for Distributive, Labor, and Social Studies at Universidad Nacional de La Plata and the World Bank (CEDLAS and The World Bank, 2023). SEDLAC is a harmonized database of socioeconomic indicators constructed from household survey microdata, applying consistent variable definitions across countries and years. It provides annual information disaggregated by urban and rural areas for 14 Latin American countries—Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El

Salvador, Honduras, Mexico, Panama, Paraguay, Peru, and Uruguay. This disaggregation is key for our analysis, as it allows us to systematically compare labor market outcomes across rural and urban areas. The dataset spans the period 2003-2023 and offers yearly information on a broad set of socioeconomic, demographic, and labor market variables. For Chile and Mexico, we complement this information with data from the LABLAC database, which follows a protocol similar to that of SEDLAC but relies on labor force surveys rather than household surveys. Table A.1 in the Appendix summarizes the availability of these sources across countries in the region. For both the descriptive analysis of labor market outcomes and the estimation of motherhood effects, we restrict the sample to individuals aged 20 to 54. To examine the evolution of rural and urban labor market conditions for women and men over the past two decades, we compute the relevant indicators for five reference years: 2003, 2008, 2013, 2018, and 2023.1 Table A.2 in the Appendix reports descriptive statistics for the main variables in the most recent year (circa 2023). In that year, the sample includes 120,830 rural women and 115,614 rural men, as well as 440,976 urban women and 380,667 urban men. In addition to labor market outcomes, we also have information on fertility-related variables. Table A.2 shows the percentage of individuals who are mothers or fathers, the age at first birth, and the number of children they have, both in rural and urban areas.2 Finally, for the analysis in section 6, we use data from Mexico’s National Time Use Survey (ENUT) for 2019, harmonized by the GenLAC project (CEDLAS, 2024), to analyze gender gaps in the time dedicated to unpaid activities, such as domestic chores and childcare, in

rural and urban areas. 1When a survey for a specific reference year was unavailable for a given country, we use the closest available year. 2These variables are defined for household head and spouse only.MARCHIONNI ET AL . 5 | Pseudo-panels In Sections 4 and 5 we estimate the causal effect of motherhood using a pseudo-event study approach (Kleven et al., 2024). While traditional event studies rely on panel data—which are scarce in the region—this approach leverages individual-level pseudo-panels constructed from cross-sectional data. Specifically, we generate these pseudo-panels based on the repeated cross-sectional surveys described in the previous subsection, closely following Marchionni and Pedrazzi (2025).3 To identify parents in our data, we focus on the heads of household and their spouses. For those already identified as parents, we estimate the calendar year of their first childbirth based on the age of their oldest child. While this method allows us to identify parents after the event, it is not possible to determine who will become a parent beforehand: crosssectional data do not provide information about whether or when childless individuals will have children. To address this limitation, Kleven (2022) suggests pairing parents with non-parents who share similar observable characteristics. For this matching process, we use variables such as age (in years), gender (male or female), education (categorized as incomplete primary, complete primary, incomplete secondary, complete secondary, incomplete tertiary, and complete tertiary), and geographic region (urban or rural).4 In addition, following Marchionni and Pedrazzi (2025), we include the survey year as a matching criterion, which allows us to pair each individual with someone from the same birth-year cohort who also shares similar observable characteristics, including gender, age, education, and region. This feature improves the matching quality

and is not feasible with census data, which are typically collected every ten years, as used in Kleven et al. (2024). Thus, household surveys provide an advantage over census data in this context. These pseudo-panels include women and men whose age at the birth of the first child is between 20 and 45 years old. The resulting sample contains 1,096,871 mothers and 1,149,647 fathers, who had children at some point before the survey takes place. Among them, 10.9% of mothers and 7.7% of fathers live in rural areas, while the rest reside in urban areas. Table A.3 in the Appendix describes the sample for the pool of countries (pooled sample). 3 | FEMALE AND RURAL: A DOUBLE DISADVANTAGE IN LABOR MARKETS Rural women in Latin America have made remarkable educational gains in recent decades. They have now surpassed rural men in years of schooling, a reversal that took place around 2013, following the urban trend with a five-year lag (see Figure A.1 in the Appendix). For instance, the share of rural women with some college education is about 5 percentage points higher than that of men, and the gap has been steadily widening over time. Despite this progress, rural women still face substantial labor market barriers. This section aims to provide a comprehensive overview of rural women’s position in the labor market in terms of labor supply and earnings, comparing them not only to rural men but also to their urban counterparts. We begin by analyzing the usual outcomes: employment, hours worked per week, hourly wages, and monthly earnings. The goal of these comparisons is to shed light on the extent to which the disadvantage faced by rural women in terms of labor supply and earnings stems from gender gaps, from rural-urban 3Unlike the descriptive analysis in Section 3, the analysis of motherhood effects in Section 4 excludes the

Dominican Republic, as the event-study estimates are extremely noisy, possibly due to the small sample size. 4For the largest countries, such as Chile, Mexico, and Colombia, the matching combines regional and urban/rural indicators.MARCHIONNI ET AL . 6 disparities, or from the intersection of both. As shown in Figure A.2 in the Appendix, labor market trends over the past two decades exhibit striking regularities: rural and urban women have followed similar trajectories across all basic labor supply indicators, as have rural and urban men. This pattern implies that rural-urban gaps have remained relatively stable over time. Part of this rural-urban gap reflects differences in women’s characteristics across areas: once we control for education and age, the employment gap decreases from 8 percentage points to just over 5. In contrast, gender gaps have narrowed substantially over time—particularly in employment, where women have seen important gains. Despite this progress, rural women continue to exhibit the lowest labor supply at both the extensive and intensive margins, underscoring a persistent structural disadvantage, as shown in Figure 1. In 2023, the employment rate for rural women stood at 56%, about eight percentage points lower than that of their urban counterparts.5 Moreover, the gender gap remains notably wider in rural areas—over 30 percentage points even after controlling for education and age—than in urban areas—22 percentage points. Figure 1 also shows that rural women not only participate less but also work fewer hours per week: in 2023, they averaged 37 hours, four hours fewer than urban women and eight hours fewer than rural men. Once again, the gender gap in hours worked is more pronounced in rural areas—8 hours even after controlling for education and age—than in urban areas—6 hours. These findings underscore that rural women face a double disadvantage in the labor market. On the one hand, they experience a persistent rurality gap characterized by lower

employment rates and fewer hours worked compared to their urban peers. On the other hand, the gender gap is also wider in rural areas, at both the extensive and intensive margins. The intersection of these two dimensions—gender and rurality—thus results in a cumulative disadvantage for rural women. While hourly wages have increased for all groups since 2003 (see Figure A.2 in the Appendix), rural women consistently earn less than rural men, who in turn earn less than both urban women and men, as shown in Figure 2. This wage gap, combined with a lower employment rate and fewer working hours, results in significantly lower total monthly earnings for rural women. In 2023, the cumulative effect of these disadvantages is evident: rural women earn 21% less than their male counterparts in rural areas and 27% less than urban women. To account for the fact that these gaps could reflect differences in age or education, we estimate conditional gender and rurality gaps. When we do so, the gender gap actually widens to 34%, suggesting a penalty for rural women relative to men beyond compositional differences, while the rurality gap among women narrows to 8%, indicating that part of their disadvantage relative to urban women is explained by observable characteristics. 4 | THE MOTHERHOOD EFFECT IN RURAL VERSUS URBAN LATIN AMERICA While rural women already face a double disadvantage in labor markets—stemming from both their gender and their place of residence—motherhood adds yet another layer of vulnerability. This section explores the additional penalty associated with motherhood in rural settings, examining how it affects women’s labor supply and earnings. 5We focus on employment because labor force participation yields similar conclusions, as shown in Figure A.2.MARCHIONNI ET AL . 7 55.688.963.485.436.844.540.546.4020406080100EmploymentHours workedper weekRural womenRural menUrban womenUrban men

(a) Levels -33.3-34.2-7.9-5.4-7.8-7.9-3.8-3.7-40-30-20-1001020EmploymentHours worked per weekGender gapRurality gapGender gapRurality gapUnconditionalConditional (b) Gender gaps and rurality gaps F I G U R E 1 Employment and weekly hours worked in Latin America, by gender and rural/urban area, circa 2023. Notes: Latin American average. The gender gaps in Panel (b) refer to rural women - rural men, while the rurality gaps refer to rural women - urban women. Employment gaps are measured in percentage points, whereas hours worked gaps are measured in weekly hours. Conditional gaps are obtained after controlling for education and age. The sample corresponds to individuals aged 20 to 54 years old. Source: authors’ own calculations based on SEDLAC (CEDLAS and The World Bank, 2023) and GenLAC (CEDLAS, 2024) datasets.MARCHIONNI ET AL . 8 3.23.33.94.24.65.86.27.702468Hourly wageMonthly earnings(in hundreds)Rural womenRural menUrban womenUrban men (a) Levels -1.8-14.3-18.50.1-21.4-34.4-27.3-7.7-40-30-20-100Hourly wageMonthly earningsGender gapRurality gapGender gapRurality gapUnconditionalConditional (b) Gender gaps and rurality gaps (in percentage) F I G U R E 2 Hourly wages and monthly earnings in Latin America, by gender and rural/urban area, circa 2023. Notes: Latin American average. In Panel (a), wages are measured in constant 2005 USD and monthly earnings in hundreds of constant 2005 USD. The gender gaps in Panel

(b) refer to rural women - rural men as a percentage of rural men’s average outcomes, while the rurality gaps refer to rural women - urban women as a percentage of urban women’s average outcomes. Conditional gaps are obtained after controlling for education and age. The sample corresponds to individuals aged 20 to 54 years old. Source: authors’ own calculations based on SEDLAC (CEDLAS and The World Bank, 2023) and GenLAC (CEDLAS, 2024) datasets. Figure 3 shows that mothers—especially those with young children—are significantly less engaged in the labor market than childless women. Although this child-related gap appears in both rural and urban areas, territorial disparities result in rural mothers exhibiting the lowest levels of employment and working hours.MARCHIONNI ET AL . 9 49.558.560.556.666.970.234.636.338.039.139.841.4020406080EmploymentHours workedper weekRural women with children 0-5Urban women with children 0-5Rural women with children 6-11Urban women with children 6-11Rural women without childrenUrban women without children F I G U R E 3 Employment and weekly hours worked for women in Latin America, by number and age of children and rural/urban area, circa 2023. Notes: Latin American average. The sample corresponds to women aged 20 to 54 years old. Source: authors’ own calculations based on SEDLAC (CEDLAS and The World Bank, 2023) and GenLAC (CEDLAS, 2024) datasets.

This raises a key question: how does motherhood affect rural women compared to their urban counterparts? The presence of children may influence labor market outcomes differently across settings due to factors such as access to childcare services, the availability of informal support networks, cultural norms regarding women’s roles, and the sectoral

composition of jobs. The relevance of this question is further amplified by the fact that rural women are more likely to be mothers, have more children on average, and start their families earlier. On average, 85% of women aged 20 to 54 in rural areas are mothers, compared to 82% in urban areas, rural mothers have 2.15 children on average, versus 1.92 among urban women, and begin their families approximately six months earlier. To answer this question, in this section we assess the causal impact of children in rural and urban settings using a pseudo-event study approach around the birth of the first child (Kleven, 2022). This method allows us to isolate the effect of the first childbirth on the labor market trajectories of mothers and fathers in both rural and urban areas. | Pseudo-event study The pseudo-event study methodology operates as an event study based on pseudo-panel data at the individual level, rather than relying on actual panel data, which are typically unavailable in most Latin American countries. Our approach closely follows Marchionni and Pedrazzi (2025), who adopt a pseudo-event study framework as in Kleven et al. (2019b) based on repeated cross-sectional data for Latin America. In this context, the event is defined as the year of the first child’s birth. Let τ represent the number of years relative to the event, where τ = 0 corresponds to the year when the first child is born.6 The structure of this event-study around the first childbirth is captured 6The timing of the first birth is inferred from the age of the oldest coresiding child. A potential concern is that, for some mothers, the actual first child may have already left the household (e.g., an older sibling who moved out), in which case the measure would capture the birth of a subsequent child. This is unlikely to occur in our setting, as we restrict the sample to mothers whose oldest coresiding child is at most 10 yearsMARCHIONNI ET AL . 10

in Equation 1: yitcτ = X k̸ =−1 βkI(k = τitc) + X j γjI(j = ageitcτ) + X y δyI(y = t) + X s λsI(s = c) +ϵitcτ, (1) where yitcτ denotes a labor market outcome for individual i observed in calendar year t, residing in country c, and at event time τ. The first term on the right-hand side includes event-time dummies, while the second and third terms capture a full set of age-in-years a