Gendered property rights and collective action for sustainable livestock systems: evidence from rangeland reform in China

Across the globe, over 1 billion people rely on grazing-based agriculture, directly and indirectly¹. Rangelands, among the most extensive terrestrial ecosystems, cover about 40% of the planet’s land surface and 69% of agricultural land². Beyond sustaining rural economies, they deliver critical ecosystem services, including habitat provision, carbon sequestration, and regulation of water and soils. Yet an estimated 49% are degraded to varying degrees, undermining these benefits and raising production risks². In this context, collective action among livestock producers is essential to sustain grazing resources and enhance system resilience. Self-governance based on collective action has long been recognized as a crucial means of mitigating environmental degradation in common-pool resources^3,4. Drawing on diverse case studies across natural resource governance, Ostrom (1990) showed that collective action organized by local communities through robust institutional arrangements can effectively sustain resource use⁵. As of 2020, at least 65 of 73 countries covering approximately 85% of the global land area had established community-based tenure regimes in law, enabling collective action for governing common-pool resources⁶. In pastoral contexts, such collective arrangements have been linked to higher biomass, greater vegetation cover, and more stable livestock productivity^7,8,9,10. Meanwhile, community grazing institutions also strengthen social-ecological resilience by lowering livestock mortality, protecting cash crops, reducing household vulnerability, and improving pasture recovery^11,12. They also raise ecological efficiency¹³, reduce greenhouse gas emissions¹⁴, and slow environmental degradation^15,16.

Despite these advances, mobilizing households to participate in sustainable collective action faces numerous challenges¹⁷. Communities are far from homogeneous units; they embody significant socio-economic inequalities and diverse interests in how they use and depend on rangelands¹⁸. Gender represents one of the most salient and persistent forms of such heterogeneity. Women and men, whether landed and landless, differ in the nature and extent of their dependence on and use of local pastures, shaped primarily by the gender division of labor and differences in economic endowments¹⁹. These constraints are deeply embedded in customary and legal institutions, with rights to resources placed in the middle because of their central influence on both motives and means and opportunities, which frequently disadvantage women^20,21,22. Even when motivated to undertake sustainable actions, a lack of resources often prevents women from doing so. Understanding such dynamics helps explain how secure land rights can strengthen women’s contribution to sustainable rangeland management and generate more inclusive social outcomes. This issue has direct relevance for ongoing policy debates around joint land titling^21,23,24, and aligns with multiple targets under the Sustainable Development Goals.

Evidence from empowerment programs shows that women adopt more cooperative strategies in natural resource decisions^25,26, and that their inclusion in governance bodies strengthens community resilience and adaptive capacity to climate risks^27,28,29. In Kerala, India, an estimated 68,000 all-women group farms demonstrate superior performance in sustaining livelihoods and resource management compared to male-led small family farms³⁰. These findings align with insights from feminist political ecology, which posits that women’s greater ecological awareness arises not from innate affinity with nature but from their differentiated labor roles and unequal property rights that heighten dependency on common resources and exposure to environmental degradation^31,32. However, the assumption that rural women are generally more conservation-oriented is not consistently supported by evidence^33,34,35,36. In some settings, greater female participation has coincided with higher forest degradation and emissions³⁷. Such mixed results likely reflect differences in asset ownership, especially land, and in resource dependence^38,39. Without secure formal rights, women’s involvement may remain superficial and fail to translate into effective collective action for sustainable management. This underscores the need to explain how institutional design and the distribution of secure rights within households influence gendered participation in collective resource governance. These insights motivate our empirical assessment of whether female titles on rangeland exert distinct effects on collective management of shared resources.

Existing quantitative literature employs three main approaches to assess the gender differences in land access and control. The first links productive efficiency to the gender of the household head or to the reported farm decision maker^22,40. The second evaluates the impact of randomized titling programs that grant equal rights or shift ownership toward women^41,42,43. The third uses plot-level comparisons within households, leveraging gendered specialization in plot management in Sub-Saharan Africa⁴⁴. While each of these approaches has yielded valuable insights, they also bear some limits. Headship and reported manager status often obscure intra-household bargaining processes and may reflect arrangements that weaken women’s claims. Titling experiments can alter de jure rights without capturing who controls inputs, labor, or output. Plot-level analyses risk conflating management roles with ownership and are less applicable to pastoral systems where collective use is common. The gap is especially significant in dryland and semi-arid regions, where rangelands dominate, and conservation challenges are acute⁴⁵. Few studies assess how women-focused titling influences collective governance outcomes. Most existing studies link collective action to social capital and connect land titling to women’s empowerment or household efficiency^46,47, implying credible pathways for our hypothesis. First, women’s peer communication can foster inclusive, trust-based groups^48,49, while formal rights recognition enhances the perceived legitimacy and willingness of titled women to participate in community affairs⁵⁰, thus improving information access, coordination, and peer monitoring⁵¹. Second, empowerment via assets and rights rebalances intra-household bargaining power, enabling women’s greater risk aversion, community orientation, and longer intertemporal preferences for sustainability to translate into household decisions⁵². Finally, secured title cultivates an identity that strengthens stewardship motivations and compliance with sustainable-use norms⁵³.

Our study examines how the allocation of rangeland rights within households influences participation in collective management. We employ the evidence from state-owned ranches in Hulun Buir, China, which, since 2003, have allocated rangeland through contractual arrangements to individual employees under a scheme referred to as the “Employee Responsibility System”. While the legal nature of these rights remains debated⁵⁴, our core focus is not on legal classification per se, but on the disparity in formalized rights between males and females, which we refer to as use rights throughout this paper. In this context, female laid-off workers’ claims to rangelands were formally recorded and recognized upon assignment based on their employment status, identical to the treatment of male laid-off workers. Drawing on data from 112 household surveys in 2024 across six production brigades in state-owned ranches in Hulun Buir, Inner Mongolia (Fig. 1), we identified the title-holder within each household, recorded participation in cooperative management, and captured household characteristics. This allows us to examine whether women who hold rangeland titles are more likely to engage in collective action and explore how these effects vary across household types and resource conditions. By integrating quantitative and qualitative evidence, we contribute to emerging debates on gendered resource governance and sustainable livestock systems.

Disconnect between women’s responsibilities and formal land rights

To contextualize our quantitative findings on collective action, we begin by exploring a central paradox observed in the division of household labor. Restricting our focus on families not dependent on hired herders, we examine the allocation of herding and decision-making roles from a gendered perspective. Figure 2 depicts the overlap of three dimensions at the household level: women exclusively hold the rangeland title, women as the primary herder, and women as the dominant livestock decision maker. Overlaps are limited: only 2 households have a woman who simultaneously exclusively hold the land title, perform herding, and make livestock decisions. Among the 112 surveyed households, 54 report a woman as the primary herder, which contrasts with the common perception that many men primarily shoulder herding responsibilities. Twenty households report women as both the primary herder and decision-maker, but without a title. Female-solely titling and female-solely decision-making cases are uncommon, while female primary herding without corresponding formal rights is frequent (51 households). An interesting phenomenon emerged repeatedly during our household surveys. In initial household interactions to identify the primary respondent, when asked directly who holds decision-making authority regarding livestock production and marketing, the response was often swift and unanimous: the husband. However, when the preliminary conversations delved into specific details, such as “What was the price per cattle at the last sale?” or “How is grazing managed at the community level?”, husbands often hesitated, seeking confirmation from their wives or admitting uncertainty. It was the wives who consistently provided accurate figures and detailed management methods. Consequently, we would then identify the wife as the primary respondent and conduct the detailed survey with her. This dynamic highlights the significant gap between the customary attribution of decision-making power to men and the knowledge-based management practices women engage in, indicating that women’s critical role in production decisions is frequently obscured by patriarchal norms⁵⁵ that designate the man as the default decision-maker. These patterns align with the observation of Batas et al.⁵⁶ that decision-making authority also follows a gendered pattern, with men controlling agricultural-related assets and women overseeing household-related assets⁵⁶. This dynamic underscores a pronounced decoupling between women’s practical responsibilities and formal property rights, motivating our empirical tests of how titling relates to collective action.

Effect of female rangeland titling on collective action

We began by estimating the Average Treatment Effect (ATE) of female-inclusive land titles on a household’s decision to participate in cooperative management. To address the key empirical concern that land title allocation may not be random and could be correlated with household characteristics that also influence cooperation, we employed a doubly robust augmented inverse-probability weights (AIPW) estimator. This method provides consistent estimates if either the model for treatment assignment or the model for the outcome is correctly specified, making it more robust than standard regression or inverse probability weighting alone. Table 1 reports our main results explaining households’ collective action participation choices, and the coefficient on female_title is positive and statistically significant across all estimation methods. Column (1) reports logit regression results with full control variables, and column (2) reports the inverse probability weighting results. Compared to these baseline estimates, the preferred AIPW specification in column (3) implies a statistically significant ATE of 0.165. Female-inclusive land titles are associated with a 16.5 percentage point increase in the probability of participating in cooperative management. The balance diagnostics are reported in Supplementary Table 2.

To illustrate the treatment effect, Fig. 3 plots the distribution of predicted probabilities for households with female titles. The solid line shows the predicted probability with the actual treatment, while the dashed line shows the counterfactual probability had female members not held the title. Predicted participation without female title clusters is concentrated in the range of 0.08–0.18, whereas with a female title is concentrated between 0.20 and 0.35. The clear rightward shift demonstrates that the treatment moves a significant mass of households from lower to higher probabilities of participation, reinforcing the substantive importance of the ATE estimate.

Disaggregated estimates by household rangeland title type

We next explore whether the effect is driven by the specific type of tenure within the household. Table 2 replaces the binary treatment variable (female_title) with a categorical variable distinguishing between male-solely, female-solely, and joint titles, and reports logit estimates of household participation in cooperative management. In columns (1) and (2), male involvement in titling, and especially male-solely titling, is associated with a statistically significant reduction in participation. Relative to cases where a woman is listed on the title, the predicted probability under male-solely titling is about 61% lower. As it is shown in column (3), using a multinomial logit for the three nominal title categories, relative to male-solely titling, joint titling is associated with higher participation. Meanwhile, female-solely titles have a positive but statistically insignificant coefficient, which is consistent with the small number of female-solely titles in our data. The results underscore the role of shared rights in promoting cooperation, in line with evidence confirmed by Oyawole et al.⁵⁷ and Lecoutere and Van Campenhout⁵⁸ that narrowing intra-household empowerment gaps raises the adoption of community-oriented and environmentally friendly practices^57,58.

Participation determinants across household characteristics

The benefits of female titling are not uniform across all households^59,60, thus we further investigated whether the treatment effect varies across resource endowments and market preferences. Figure 4 profiles the heterogeneity in the ATE across subgroups defined by key household characteristics and shows that the gains from titling are largest for households that have lower income, have weaker social networks, and are actively renting in land. Whereas Auer et al.⁶¹ emphasize high social capital as the primary driver of collective action⁶¹, a plausible mechanism of our result is that these households face higher coordination needs, higher dependence on grazing system, and greater exposure to shocks, so the expected returns to cooperation are higher and the insurance value of collective management is greater⁶². This interpretation accords with classic studies of Baland and Platteau⁶³ and Wade⁶⁴ on village cooperation under risk and scarcity^64,63. This pattern underscores that titling programs that include women can be most consequential where constraints are tight and risk is salient.

Tracing the pathway to natural resource governance outcomes

Finally, we provide preliminary evidence on the potential ecological payoff of the increased collective actions spurred by women’s land rights. Having shown that including women as use right-holders is associated with higher participation in cooperative management, we ask whether greater participation improves ecological and economic outcomes. Figure 5 reports the correlation of the overgrazing ratio and household livestock net income with cooperative management. Conditional on exceeding the theoretical overstocking rate, non-cooperative households exhibited greater variability and several extreme overstocking cases, whereas cooperative management households showed a stable grazing pressure. Households participating in cooperative management exhibit significantly higher livestock net income, with an estimated 30 log-point, or approximately 35% increase relative to non-participants (p = 0.011). These results support the mechanism in which women’s inclusion in rangeland titles raises collective management participation, which in turn leads to better natural resource governance outcomes.

Rights recognition, participation, and tenure expectations

Qualitative evidence from structured surveys and in-depth interviews clarifies the channels through which women’s rangeland rights translate into collective action participation. Our further analysis of interviews with a subsample of 59 female respondents identifies mediating mechanisms reflected in responses that differ from those of their counterparts. We argue that female-led titling in rangeland rights increases household participation in cooperative natural resource governance through three channels: higher engagement in community affairs, stronger perceived tenure security, and longer expected right durations. The patterns below are consistent with the quantitative estimates and complement them with household decision evidence. First, women recorded as right-holders report higher community affairs engagement. On average, they attend community meetings 3.05 times per year, compared with 2.84 times among female respondents without rights. This difference accords with the view of Grillos (2018) that recognized decision-making rights strengthen responsibility, which in turn increases participation in collective governance⁶⁵. Enhanced representation at the village level may also improve access to information and benefits.

Second, perceptions of tenure risk also differ sharply. Among female respondents without rights (n = 35), 65.7% anticipated that their household’s land could be reallocated through policy adjustments such as expropriation. Among women with rights (n = 24), only 33.3% mentioned policy risk; few respondents linked security to employment status (n = 6) or to marital change (n = 2). The contrast indicates that formal recognition lowers perceived policy uncertainty and supports a more stable expectation of control rights, thereby enhancing willingness to invest in rangeland, which is consistent with the finding of Dillon and Voena (2018)⁶⁶.

Third, expectations about future access exhibit the same pattern. Women without rights expected to continue using their current rangelands for 25.6 years on average, compared with 37 years among female right holders, indicating a longer expected duration strengthens incentives for long-term investment and coordinated grazing, consistent with common-pool resource theory⁵. A woman in a jointly titled household told us:

“Degradation of our pastures is severe. We must follow the spring grazing ban, but when others do not, their noncompliance imposes costs on us, so the challenge is beyond any single family. I proposed allocating pastures by livestock type and season, with regular rotation, to reduce trampling and avoid concentrated browsing.”

Her awareness links perceived security to a preference for coordinated management. She added: “Some of the village members still occasionally violate the agreement,” which indicates the need for stronger formal institutions and monitoring to sustain collective action.

Consistent with prior work, female right-holders place greater weight on risk management, collective welfare, and long-term ecological sustainability than the male right-holders typically observed in the literature^67,68. Empowerment therefore does more than raise perceived tenure security and initiative, and it brings these preferences into resource decisions. When women hold secure rights and participate in natural resource governance, they tend to shift toward cooperation.

In our case, the Employee Responsibility System registered rangeland at the individual level for laid-off employees rather than at the household level. This approach allowed female workers’ claims to be visible in records and enforceable in practice as well as providing a unique opportunity to investigate the gender difference in common-pool resource management. Drawing on a unique institutional setting that isolates the effect of gendered land rights from household-level confounders, we employ a doubly robust estimation strategy to show that women’s land tenure increases cooperative management participation by 16.5 percentage points. This finding complements ongoing discourse on women’s role in natural resource management by demonstrating that empowering women through formal, individual rights creates a foundation for their participation and leadership in broader collective endeavors for rangeland sustainability.

While the previous body of literature has offered limited quantitative or qualitative evidence regarding the domain of land tenure, debates around where women-targeted interventions would yield higher returns on natural resource governance have driven a series of policy experiments and discussions rethinking representation in user groups. Case studies consistently show that when women are involved in natural resource governance, they can take the lead in long-term conservation efforts. Examples include the Chipko movement in India, lab-in-the-field experiment in Indonesia, Peru, and Tanzania, and the forest co-management program in Malawi and China^53,69,70, all of which typically apply household head titles or quotas for women on resource governance committees to represent women’s empowerment. Given the widespread reality in developing regions that women’s participation in land management exceeds their formal ownership shares (details in Supplementary Table 5), we bring a land tenure lens to the question of how to strengthen women’s roles in collective governance of common-pool resources and address the fundamental mismatch between rights and responsibilities.

Our findings highlight that legally recognized rights provide institutional security necessary for women to exert meaningful influence over decisions regarding common-pool resources and shape broader governance outcomes. It suggests that having a woman’s name on a land contract can shift household decision-making dynamics towards more cooperative and sustainable outcomes. This finding aligns with results reported by Agarwal⁵³, Cook et al.,⁶⁹ and Mazunda and Shively⁷⁰ in terms of forest governance^53,69,70, which demonstrates that greater involvement of women in community institutions leads to better conservation. However, we identify a prior, more foundational step: the formal recognition of rights at the household level. We show that before women can effectively participate in community-level collective action, they often first need the intra-household legitimacy that potential comes with legally recognized individual property rights to the land. The mixed evidence in the literature, where women’s involvement sometimes correlates with resource degradation^34,36,37, may stem from a failure to distinguish between mere participation and empowered participation. Studies using female headship or lab-in-the-field games as proxies for empowerment capture aspects of agency but often miss the foundational role of secure, formal land property rights. In contrast, our measure of female-inclusive titling captures a concrete, transferable, and socially recognized right that fundamentally changes a woman’s long-term incentives for sustainable stewardship.

Our findings place women’s leadership in collective action at the center of strategies for resilience and sustainable livestock systems, especially under the ongoing feminization of agriculture and pastoralism, where increased female labor has not automatically translated into empowerment. But we do not essentialize women (or men) as inherently conservation-minded. Rather, we underscore that effective governance depends on recognizing both women’s and men’s preferences, their intangible and intrinsic motivations, and their material conditions and means. The stronger effects we find among joint-titling households further suggest that it is not about transferring control from men to women, but rather creating a foundation for shared, cooperative stewardship within the family unit. This perspective moves the policy debate beyond a simplistic binary of “male” vs. “female” control towards a more sophisticated understanding of formalized shared rights and highlights female de jure claims as a key pathway to collective action. Conversely, the result that sole male tenure is associated with a lower likelihood of cooperation challenges the traditional patriarchal assumptions⁵⁵, revealing that the default assignment of rights to men may inadvertently hinder sustainable rangeland management.

Placing these findings in the wider context of the ongoing feminization of agriculture⁷¹, there is a persistent misconception that women farmers are less productive than their male counterparts, a view often rooted in women’s more limited access to assets and services rather than in effort or ability⁷². In reality, women are actively involved in essential livestock system tasks such as fodder harvesting, by-product food processing, grazing and livestock raising, dung cake making, animal selling, marketing strategy determination, milking, and managing fallows^73,74,75. Yet their formal rights rarely match these responsibilities (details in Supplementary Table 5). A common thread, despite contextual variation persists across Sub-Saharan Africa, South Asia, Latin America, and China, is that formal commitments to equality frequently coexist with customary norms, elite control, and procedural barriers that weaken women’s claims^{76,77,78,79,80}. Even as gender equality has moved into policy agendas and reforms globally, the rights accorded to women have not kept pace with the responsibilities they shoulder in agriculture. The manifestations of this disparity differ by context. In Sub-Saharan Africa and South Asia, conflicts between statutory law and customary practice restrict women’s registration and inheritance, with ethnicity often reinforcing unequal control over land.

Relatedly, a common ground of research specific to China and Latin America is that land is often mediated by community membership, which can systematically disadvantage women^81,82. The third National Survey on the Social Status of Women in China indicates that 27.7% of women lose land following marital changes such as marriage, remarriage, divorce, and widowhood, often because they simultaneously lose their collective community membership. In 2010, the proportion of landless rural women was 9.1% higher than that of men⁸³. Although the 2018 amendment to the Organic Law of the Villagers’ Committees of China prescribes that “female representatives should constitute no less than one-third of the total membership in the villagers’ representative assembly”, monitoring data indicate that women made up only 23.1% of village committee members in 2017⁸³. Within the village self-governance system, customary practices and local procedures can therefore erode women’s tenure security and limit formal complaint channels when collective organizations infringe on members’ rights. However, our case from China’s state ranch system presents a critical divergence through the absence of traditional village self-governance bodies. Due to the historical legacy, village self-governance committees were not established in our study area, and government oversight within state-ranch jurisdictions remains limited. This unique administrative structure allows us to isolate the effect of intra-household land rights from the confounding role of community-level hierarchies typically found in other rural settings.

Our findings are based on a specific and relatively small sample, which warrant caution in generalizing these results to broader contexts. Our empirical evidence speak to a tendency to overlook both women’s dependence on natural resources and their roles in managing them. Understanding these dynamics requires looking beyond ownership and examining the distribution of decision-making power and bargaining power within households and communities. These factors fundamentally influence how men and women can act on their motivations to conserve resources. Furthermore, our sample size limits the statistical power to quantitatively evaluate complex mediating effects. Consequently, the mechanism we proposed, linking land titles to empowerment, decision making, and cooperative governance, should be interpreted as a theoretically grounded pathway supported by qualitative evidence rather than an established causal assertion. Our findings suggest that individual titling does not automatically translate into empowerment and an immediate elevation of social status. Instead, formalized tenure provides a crucial institutional pathway that enhances the recognition of women’s labor and property rights, thereby amplifying their voice and representative capacity in local resource governance. We also acknowledge that our survey did not systematically interview husbands and wives separately, future studies should benefit from larger-scale surveys integrating standardized empowerment instruments and dual-respondent surveys to enable quantitative validation of these underlying mechanisms. To translate these insights into action, local common-pool resource tenure systems should formally recognize women’s land rights. Governance systems must establish transparent rules and monitoring mechanisms to link resource rights clearly with responsibilities, as well as support collective actions. It is equally essential to ensure women’s participation in resource management committees and to provide agricultural extension services that actively engage both spouses. Capacity enhancement schemes through training, extension, and the formation of farmer associations and women’s cooperatives can raise skills, build networks, and improve access to finance and other physical assets. Such measures are mutually reinforcing with land titling legal reform. Finally, we call for a broader shift in evaluations on gendered issues that capture social and ecological sustainability, rather than short-term productivity measures.

Study area and institutional context

The Hulun Buir municipal state-owned ranch located in Inner Mongolia was selected as the research site because it is one of the earliest state-owned ranches to implement property rights reform in China. Facing the pressure of low managerial efficiency and surplus staffing, employee land responsibility pilots that allowed employees to possess the use rights of the state and take full responsibility for their profits and losses began in 1979 in the context of state enterprise reform and the implementation of the rural household responsibility system, and received central approval in 1983^84,85,86. The ranch manages about 1.1 million hectares, including 667,000 hectares of rangelands and 40,000 hectares of cropland. It ranks third in terms of cropland scale among state-owned farms(ranches) nationwide and is a leader in the development of animal husbandry in the Hulun Buir region. The vegetation ranges from meadow steppe to typical steppe and provides critical ecological services. The annual average potential evapotranspiration is 2250 millimeters, consistent with a semiarid grassland climate⁸⁷. Given the extensive rangelands and unique natural conditions, local employees predominantly raise cattle, sheep, goats, and horses for their livelihood.

Since 2003, through several rounds of adjustment, the Hulun Buir state ranch has consolidated the Employee Responsibility System and assigned state rangeland use rights to laid-off employees, allowing eligible employees to forgo wages in exchange for a rangeland contract registered in the individual’s name rather than the household. At that time, tenure contracts were allocated directly to laid-off state employees of both genders based strictly on their employment status. Typical entitlements included 16.67 hectares of hayfield and 10 hectares of grazing land. The inheritance and transfer of state-owned agricultural land use rights are prohibited. While minor spatial administrative adjustments may occur, rights are formally revoked or reallocated mostly upon the retirement of these former employees. This institutional arrangement provides plausibly exogenous variation in land-use rights, employment status at the time of reform may still correlate with certain household characteristics. We therefore interpret our identification as leveraging historically determined variation within a common institutional setting, rather than assuming full exogeneity.

Data collection

Fieldwork was conducted in 2024 across six neighboring production brigades within the state ranch. These brigades share similar institutional, climatic, and geomorphological contexts, thereby providing a set of field-based comparative cases. We use the term “production brigade” to denote the smallest organizational unit of state ranches inherited from the collectivization era in China. Each brigade functions as a village-like settlement of state ranch employees and their families. Initially, state ranches also handled a full set of public administrative functions, but de-administration reforms in the 1990s shifted their emphasis to economic management. Consequently, most brigade affairs are now managed through self-governance with limited external government intervention. Within the six sample brigades, we conducted a census of all accessible households that had received rangeland tenure during the 2003 institutional reform.

All respondents are former workers who directly participated in the 2003 reform, currently hold permanent residence, and rely primarily on livestock income. We interviewed all households at home in the selected brigades, supplemented by telephone interviews for individuals who were not home, ensuring the dataset captures all accessible households in each study brigade. This process yielded a total of 112 valid responses, including 61 from women, 46 from men, and 5 where the respondent’s gender was not recorded. It is worth noting that in households with a female respondent, 34.43% had a female title holder, compared to 26.09% in households with a male respondent, indicating a relatively balanced distribution of female title holders across both respondent genders.

The survey collected data on household demographics and socioeconomic characteristics, livestock management practices, as well as perceptions over rangeland condition and land regime. Specifically, we utilized recall data from 2023 because it represents the most recent complete livestock production cycle. To minimize potential bias and ensure the quality of responses, we followed a specific interview protocol. Upon visiting a household, we first identified which member was most knowledgeable about livestock management. While both spouses might initially greet the interviewer, the detailed questionnaire including individual perceptions and tenure security was conducted with the identified respondent individually. This approach prevented answers from being influenced by the presence of others and allowed us to capture the genuine perspectives of those most involved in rangeland governance, and to record the interaction patterns between household members.

Table 3 lists the descriptive variables of households by rangeland title distribution. There are slight differences in the number of observations, as not all the respondents answered every question in the survey. Complete-case data are available for 112 observations, which are classified into three household types based on intra-household title arrangements, including 75 male-solely titled households, 5 female-solely titled households, and 32 joint titled households, where joint title denotes spousal joint registration on the rangeland use right contract. Descriptive results indicate that joint-titling households exhibit higher rates of cooperative management, larger herd sizes, and lower loan burdens compared to their solely titled counterparts. They also maintain the largest herds on average, with considerable dispersion. Specifically, joint-titled households or female-titled households are more likely to engage in cooperative operation than male-titled households. Moreover, where cooperation occurs, the scale of cooperation is much larger among joint-titled households. Although female-titled households report the highest livestock income (CNY 178,400), they also experience the highest incidence of outstanding loans (40%) and possess the smallest average livestock-shed area (148.33 m²). This pattern suggests a greater reliance on credit to sustain or expand herd, consistent with constraints on asset acquisition, capital accumulation, and network access to inputs when women hold titles solely. These descriptive results underscore the necessity of clarifying potential selection biases.

Ethical considerations

This study involved voluntary household interviews with adult participants and was performed in accordance with the Declaration of Helsinki. As this research was a non-clinical, observational social science survey, formal ethical approval and a specific reference number from an Institutional Review Board (IRB) were not required under institutional guidelines at the time of data collection. All participants were informed of the research’s academic purpose, their right to decline to answer any question, and their right to withdraw at any time. Participation was voluntary, and oral informed consent was obtained before each interview. To protect participant privacy, the raw survey data are kept confidential. All analyses and results presented in this research are based on aggregated data, and no personally identifiable information is used or disclosed.

Variable definition

The outcome variable is household participation in collective action, measured by a binary variable Cooperative management that equals 1 if the household reports taking part in a collective arrangement in 2023 and 0 otherwise. We also inquired about the management strategy in 2024, and the household participation rate was 16.67% in 2023 and 15.74% in 2024 in the total sample, and the scale of existing cooperation remained unchanged across both years, which is relatively stable. Specifically, in this study, cooperative management refers to a reciprocity-based arrangement conducted jointly with one or more partner households. Practices include labor sharing for daily herding, seasonal mobility, watering, enclosure management, boundary patrols, jointly designating grazing zones and rotating their use, and related rangeland management tasks, which were common while declining in pastoral communities of China⁸⁸. The key independent variable is female title, which equals 1 if at least one adult woman is listed on the rangeland use-right certificate (female-solely or joint title) and 0 if the title is male-solely owned; in robustness checks we replace this binary with a nominal variable distinguishing male-solely, female-solely, and joint titles.

Learning from previous studies^80,89 as well as to limit post-treatment bias, we include only covariates that could plausibly affect both the selection into the Employee Responsibility System reform and intra-household allocation of rights and the participation in collective action, while excluding mediator variables that are likely consequences of titling, such as contracted land area, herd size, current cooperative membership, and household income. In the baseline specification we therefore control gender of household head, as female land title may be related to female household headship and the absence of adult males²², we separate women’s registration on the use right from households without an adult male, and capture differences in labor availability and bargaining structure. Cadre connections, indicator equal to 1 if any immediate relative holds a managerial position in government or a state-owned enterprise, capturing social networks and access to information⁸⁹. We also controlled age and years of education for the title holder. For jointly registered certificates, we use the household head’s attributes, since these variables proxy human capital and may influence intra-household decision making.

To evaluate the potential ecological payoff of collective actions, we calculated the overgrazing ratio, which compares actual stocking ((Y)) to theoretical capacity ((T)). Specifically, the actual stocking rate (Y) represents the net grazing pressure per hectare of rangeland area ((A)), calculated by converting year-end livestock inventory (({Q}_{i})) after sales (({S}_{i})) into standardized sheep units (SU) using conversion coefficients (({gamma }_{i})) with full details provided in Supplementary Table 4. Theoretical carrying capacity (T) is defined as the maximum sustainable number of SU that a given pasture unit can support during the grazing season, derived from the balance between available natural forage and basic animal requirements.

Where the (mathrm{NPP}) in Eq. (3) is the net primary productivity ((mathrm{kg}{rm{C}},{{rm{m}}}^{-2},{mathrm{year}}^{-1})) extracted within township-level rangeland polygons from MODIS17A3 dataset⁹⁰. Since grazers only consume above-ground biomass, we apply a below-ground NPP fraction (({f}_{{bnpp}})) of 0.698⁹¹. The utilization coefficient (k), set to 0.55, represents the proportion of available forage actually consumed by livestock during the graze, guided by NY/T 635–2015 issued by the Ministry of Agriculture of China⁹². To ensure consistent measurement units across all biomass data, animal demand is expressed in carbon mass. Specifically, the daily feed intake per sheep unit ((M)) is set to 0.697 kg C·({mathrm{SU}}^{-1})·({mathrm{day}}^{-1}), which is equivalent to 1.548 kg dry weight multiplied by a 0.45 carbon conversion coefficient⁹³. Finally, the annual grazing day ((D)) is set to approximately 180 days based on local grazing practices observed during field surveys.

Empirical estimation

We analyze land title as a treatment policy and household participation in cooperative management in 2023 as an outcome to study the impact of gendered land title. A key empirical challenge is the potential endogeneity and self-selection in land allocation. Households in which the female spouse was the state-owned ranch employee may differ systematically in education, social capital, leadership ties, or network access, all of which can raise the likelihood that the woman is listed on the title and participation in collective action. To address this concern, we estimate the average treatment effect (ATE) using the doubly robust method, as it combines reweighting (through an inverse-probability regression) and regression adjustment to estimate the treatment effects and provides parameter robustness. This method has been employed by a range of studies assessing the effect of intra-household empowerment^94,95. Compared with inverse-probability weighted regression adjustment (IPWRA), augmented inverse-probability weights (AIPW) offer greater robustness to model misspecification by incorporating a bias-correction term. AIPW specifies outcome model (cooperative management model) and treatment model (land title model) together. Treatment model is used to match counterfactual observations and reweight outcome result for avoiding self-selection and wrong specifications of the outcome model. Bias-corrected term in AIPW is arranged to avoid wrong specifications of the treatment model.

The treatment model is specified as a logistic regression:

Where (varLambda left(cdot right)) is the logistic cumulative distribution function, ({D}_{i}) indicates whether household i holds a female land title, and ({X}_{i}) is a vector of covariates.

The outcome models for cooperative management participation are specified as:

The AIPW estimator of the ATE is

This estimator is doubly robust: consistency obtains if either the propensity-score model or the outcome models are correctly specified.