Effect of Sustainable Livelihood Capital on the Productive and Operational Activities of the Female Labor Force in Vietnam’s Mangrove Areas

Abstract

This study examines the impact of Sustainable Livelihood Capital (SLC) on women’s participation in production activities in Vietnam’s mangrove areas, using a gender-focused approach to advance gender-sensitive livelihood theory. Employing binary Logit regression and cross-tabulation analysis, model robustness was confirmed via the Bayesian Information Criterion (BIC). Findings identify financial and human capital as core factors influencing participation, with access to credit emerging as the factor exhibiting the strongest correlation. Livelihood stability and practical vocational training support women’s long-term productive engagement. The study highlights the livelihood paradox and resource lock-in effects: over-reliance on mangrove income reduces participation, limiting diversification, while over-exploitation correlates positively with participation as a survival strategy, exerting short-term environmental pressure. Conversely, owning traditional assets like fishing boats negatively affects participation, showing traditional factors hinder economic restructuring. The findings of heterogeneity analysis emphasize the necessity of policy intervention. For example, women under 56 are primarily financially driven; those over 56 exhibit lower participation and face severe human capital bottlenecks, especially education. Larger households face significant financial barriers despite having abundant human capital. Married women face dual constraints from financial and traditional physical capital, while single/divorced women hold advantages in education and opportunities. Furthermore, In areas far from fishing ports, financial and human capital are core drivers. This research provides quantitative evidence on the complex, heterogeneous effects of SLC on women’s productive engagement, offering a scientific foundation for multi-dimensional, targeted policy measures to foster sustainable livelihood diversification.

1. Introduction

The mangrove forests in the Ca Mau region are not only the largest ecosystem in Vietnam but also the foundation of livelihoods for local communities [1]. The female labor force, in particular, plays a significant role in productive and operational activities, being the main participants in various sectors ranging from aquaculture and processing to ecotourism [2]. However, the female labor force in Vietnam’s mangrove areas faces serious livelihood obstacles, including a lack of sustainable livelihood capital and external challenges posed by climate change and resource depletion [3]. These constraints reduce women’s capacity to participate in high-value and sustainable economic activities.

Theoretically, although the multi-dimensional livelihood analysis framework (DFID) has clarified the central role of five types of capital (social, physical, natural, financial, and human) in shaping livelihood strategies [4], significant gaps remain in research on the systematic interactions between these capitals. Most existing studies remain at the level of describing single factors. In particular, current research has failed to adequately systematize gender factors, overlooking the inequalities faced by the female labor force in accessing and controlling resources, which directly affects their ability to participate in sustainable livelihood pathways [5].

To address the limitations of existing research, this paper focuses on gender inequality in accessing livelihood capital within mangrove coastal communities, particularly differences between generations. Specifically, this paper employs a binary Logit regression model, cross-tabulation data analysis, and stability tests based on the BIC to analyze the decision-making factors through which sustainable livelihood capital influences the participation of the female labor force in productive and operational activities. This systematic analytical framework not only identifies the key types of capital affecting the participation of the female labor force in such activities but also provides a theoretical basis for formulating gender-sensitive livelihood policies. Compared to existing studies, the marginal contributions of this paper are mainly as follows: First, by systematically integrating gender factors into the SLC analytical framework, this paper makes an important theoretical contribution to the conceptual expansion and deepening of sustainable livelihoods, advancing the discourse on gender-sensitive livelihoods. Second, this paper provides empirical evidence on gender disparities in livelihood capital access and utilization, thereby contributing to an examination of the role of specific capital forms—such as financial and human capital—as well as the association between reliance on traditional resources and women’s labor force participation. Third, this paper provides empirical evidence that may inform discussions on multi-dimensional, gender-sensitive policy interventions. The findings suggest potential directions for supporting livelihood diversification and strengthening women’s adaptive capacity in the context of climate change and inclusive development, although further research would be needed to confirm policy applicability. To substantiate the aforementioned theoretical and practical contributions, this study examines the livelihood factors shaping women’s labor force participation in productive activities and proposes actionable strategies to advance sustainable livelihoods.

2. Theoretical Analysis and Hypotheses Development

Sustainable livelihoods refer to the capacity of rural households to cope with, recover from, and maintain or enhance their capabilities and assets under external pressures without degrading natural resources, thereby achieving a positive cycle between livelihoods and sustainable development [4]. The concept emphasizes balancing immediate survival with long-term ecological integrity [6]. Sustainability depends on how households access and allocate capital under the influence of various policies and institutions [7]. The Sustainable Livelihoods Framework (SLF) serves as a tool to identify drivers of poverty and enhance income and stability [8,9], while the Sustainable Livelihoods Approach (SLA) emphasizes the integration of capital, strategies, and outcomes, promoting risk resilience through diversification [10,11]. Together, these approaches provide a comprehensive perspective on poverty and vulnerability, forming the basis for effective policy interventions. The five basic types of capital include:

(1)

Social capital is essential for sustainable rural livelihoods [12,13]. In Vietnamese coastal communities, cultural values and social cohesion underpin stability, as networks based on trust and reciprocity facilitate collective action and natural resource management [14,15]. Social capital acts as an endogenous driving force for rural sustainable development, although disruptions may create livelihood risks [16]. Social networks and participation in business entities support transitions toward non-agricultural strategies [4,17]. The extent of trust and social networks strongly influences household livelihoods [10,18]. However, systemic gender barriers limit women’s access to digital, financial, and public economic resources [19,20].

H1: 

Social capital has a significant positive impact on the likelihood of female labor participating in productive and operational activities in mangrove areas.

(2)

Physical capital includes infrastructure and production means that households use to sustain their livelihoods [21]. Abundant physical assets enhance investment capacity and resilience to market fluctuations or disasters [22]. Infrastructure and household assets reflect changes in physical capital [23]. However, physical capital may have dual effects: it can lead to unsustainable land-use practices in rural tourism development [24,25] while also improving carbon emission efficiency on cultivated land [26]. These findings indicate that physical capital plays a complex, multidimensional role in livelihood strategies and sustainable development.

H2: 

Physical capital has a significant negative impact on the likelihood of female labor participating in productive and operational activities in mangrove areas.

(3)

Human capital reflects the quantity and quality of agricultural labor [27]. Farmers with higher human capital typically possess richer knowledge and skills [28]. Human capital promotes the adoption of green production technologies, especially during crop growth [29]. Accumulation through education, training, and health security strengthens skills, innovation, and economic growth [30]. Therefore, human capital is central to sustainable livelihood strategies and rural transformation.

H3: 

Human capital has a significant positive impact on the likelihood of female labor participating in productive and operational activities in mangrove areas.

(4)

Financial capital plays a key role in promoting agricultural and forestry development and enhancing farmers’ income [31,32]. Agricultural investment depends on available financial resources [33]. Inflows of financial capital into rural areas promote new economic entities, expand value chains, and diversify production [34,35]. Income from timber, livestock, and ecotourism also strengthens financial capital [36].

H4: 

Financial capital a multidimensional impact on the likelihood of female labor participating in productive and operational activities in mangrove areas.

(5)

Natural capital is key in women’s livelihoods, with strategies shifting from reliance on natural resources to engagement in service industries [37]. Furthermore, studies emphasize the crucial role of natural resources in determining livelihood stability and transition pathways, especially in areas surrounding national parks or ecologically sensitive regions [38]. Therefore, this study proposes the following hypothesis:

H5: 

Natural capital has a significant negative impact on the likelihood of the female labor force participating in productive and operational activities in mangrove areas.

3. Methodology

3.1. Data Source

This study is based primarily on field survey data collected in the mangrove areas of Ca Mau Province, Vietnam, between July and August 2025. The survey covered several districts with relatively large mangrove areas, including Ngoc Hien, Nam Can, Dam Doi, Phu Tan, Tran Van Thoi, and U Minh. Using a stratified random sampling method, 256 questionnaires were distributed and all were recovered, of which 250 were valid, yielding an effective response rate of 97.66%. The respondents were female laborers from different regions and age groups who actively participate in economic activities closely related to mangrove resources, such as fishing and aquaculture, aquatic product processing, and ecotourism. These activities are widely practiced in coastal communities and depend heavily on natural resources, particularly the mangrove ecosystem. The study analyzes the relationship between five types of livelihood capital and the participation of female laborers in productive and operational activities. These include natural capital (mangrove area, natural disasters and climate change, water quality, and saltwater intrusion), social capital (gender bias and employment training), physical capital (infrastructure, production equipment, transportation systems, and irrigation facilities), financial capital (access to funds, household income, and income diversity), and human capital (labor participation, education level, and health conditions). Data were collected through face-to-face questionnaires and on-site interviews, then processed using Excel and Stata 17.0. A binary Logit regression model is applied to examine how these factors influence women’s participation in economic activities in mangrove areas.

3.2. Variables

In this study, the dependent variable is the participation status of the female labor force in productive and operational activities. This variable is coded as a binary variable, set to 1 when the female labor force participates in productive and operational activities, and set to 0 otherwise. Specifically, within the research sample, 104 individuals did not participate (coded as 0), accounting for 43% of the total sample; 146 individuals participated (coded as 1), accounting for 57% of the total sample. This variable reflects the basic situation of the female labor force’s participation in productive and operational activities in the mangrove areas.

3.3. Methods

3.3.1. Bayesian Information Criterion (BIC)

The Bayesian Information Criterion (BIC) is primarily used to select the optimal model when comparing different models, especially when the observed sample size is limited [39]

BIC = k ln(n) − 2 ln(L)

(1)

where k is the number of parameters in the model, n is the sample size, and L is the maximum likelihood value of the model. −2 ln(L) represents the part measuring the fit of the model to the data (the smaller this value, the better the model fit). The term k ln(n) is a penalty term, reflecting the complexity of the model. When the model has more parameters, or when the data volume increases, the value of the penalty term also increases.

3.3.2. Binary Logit Regression Model

For cases where the dependent variable is binary data, Logit and Probit models should be used, as these two models can ensure that the predicted probability values always lie within the reasonable range [0, 1] [40]. Unlike OLS, which may produce unreasonable values, Logit and Probit use distributions to ensure that the predicted probabilities are always between 0 and 1. Therefore, these two models are more appropriate for dealing with probability problems involving binary dependent variables. In this study, the dependent variable is whether the female labor force participates in productive and operational activities. This variable is a binary dummy variable, meeting the requirements of the model. Hence, this paper adopts the binary Logit regression model to analyze the impact of sustainable livelihood capital on the participation of the female labor force in productive and operational activities. Specifically, the sustainable livelihood capital factors, including natural capital, social capital, human capital, financial capital, and physical capital, are used as independent variables. The specific formula is as follows:

$$\begin{array}{cc}\hfill logit\left(prodpart\right)& ={\beta }_0+{\beta }_{1}age\text{\_}56plus+{\beta }_{2}hh\text{\_}large+{\beta }_{3}marital\hfill \\ & +{\beta }_{4}near\text{\_}fish\text{\_}port+{\beta }_{5}edu\text{\_}high+{\beta }_{6}gender\text{\_}div\hfill \\ & +{\beta }_{7}skill\text{\_}income+{\beta }_{8}female\text{\_}labor+{\beta }_{9}income\text{\_}stable\hfill \\ & +{\beta }_{10}training\text{\_}effect+{\beta }_{11}boat\text{\_}asset\hfill \\ & +{\beta }_{12}waterway\text{\_}improve+{\beta }_{13}credit\text{\_}access\hfill \\ & +{\beta }_{14}income\text{\_}share+{\beta }_{15}over\text{\_}+\epsilon \hfill \end{array}$$

(2)

In the processing, if the female labor force participates in productive and operational activities, the dependent variable prod_part is assigned a value of 1; if not, it is assigned 0. The parameters ${\mathrm{\beta }}_0,{\mathrm{\beta }}_1,\dots ,{\mathrm{\beta }}_{\mathrm{i}}$ are the coefficients to be estimated, while age_56plus, hh_large, …, over_logging are the independent variables (Table 1). In the calculation results, the degree of influence of each indicator can be determined by analyzing the significance level (Sig), regression coefficient (β), and odds ratio (OR). If the significance level is less than 0.1, 0.05, or 0.01, and the regression coefficient is positive, it indicates that the variable has a significant positive impact on the participation of the female labor force in productive and operational activities. Conversely, if the regression coefficient is negative, the variable has a negative impact on the participation of the female labor force in productive and operational activities.

Table 1. Variable definitions, measurements, and descriptive statistics.

Variables		Definition	Measurement	Mean	S.D.
Dependent variable	prod_part	Whether to participate in production and business activities	1 = Yes, 0 = No	0.584	0.494
Control variables	age_56plus	Are you aged 56 years or older?	1 = Yes, 0 = No	0.16	0.367
	hh_large	Household size of 6 persons or more	1 = Yes, 0 = No	0.208	0.407
	marital
	Single	Marital status	1 = Single; 2 = Married; 3 = Divorced; 4 = Widowed	0.1	0.301
	Married			0.788	0.41
	Divorced			0.072	0.259
	Widowed			0.04	0.196
	near_fish_port	Whether the household is located near fishing ports, boat docking sites, or tidal-flat fishing areas	1 = Yes, 0 = No	0.128	0.335
Social capital	gender_div
	female led	The gender-based allocation of labor	1 = Women occupy low-income positions;	0.328	0.47
	gender equal		2 = Equal distribution between men and women;	0.14	0.348
	male important		3 = Men occupy more important positions;	0.5	0.501
	Minimal division		4 = No significant difference	0.032	0.176
	skill_income	Whether possessing skills increases income
	Def increase		1 = Positive; 2 = Dependent on opportunity; 3 = Uncertain (labor market instability); 4 = No impact	0.196	0.398
	Prob increase			0.484	0.501
	Uncertain			0.248	0.433
	No increase			0.072	0.259
Human capital	edu_high	Whether the respondent has completed high school or above	1 = Completed high school or above; 0 = otherwise	0.324	0.469
	female_labor	Number of female laborers in the household (aged 15 and above)		1.772	0.974
	income_stable	Whether employment in mangrove areas ensures livelihood stability	1 = Yes, 0 = No	0.724	0.448
	training_effective	Whether advanced or specialized training contributes to job performance	1 = Yes, 0 = No	0.708	0.456
Physical capital	boat_asset	Whether the household owns fishing boats or production equipment	1 = Yes, 0 = No	0.248	0.433
	waterway_infra	Whether improvements in waterway transportation facilitate the development of production activities	1 = Yes, 0 = No	0.928	0.259
Financial capital	credit_access	Whether it is easy to obtain loans for mangrove-related economic activities	1 = Yes, 0 = No	0.948	0.222
	income_share	Share of income from mangrove-related activities
	Below 25%		1 = Below 25%; 2 = 25%–50%;	0.432	0.496
	25%–50%		3 = 50%–75%; 4 = Above 75%	0.1	0.301
	50%–75%			0.156	0.364
	Above 75%			0.312	0.464
Natural capital	over_logging	Whether mangrove forests are over-logged	1 = Yes, 0 = No	0.164	0.371

4. Results

4.1. Baseline Regression

This paper employs a binary Logit regression model to explore the impact of sustainable human capital on the participation of the female labor force in productive and operational activities. The regression results are shown in Table 2. Among them, Model M2 is the full model, incorporating all explanatory variables from each dimension of sustainable livelihood capital, including education level, ownership of boat assets, access to credit for economic activities related to mangroves, perception of the enhancing role of skills training, while also controlling for control variables such as age (over 56), household size (over 6 people), marital status, and geographical location. Model M1 is a simplified model, retaining only the core independent variables and excluding the aforementioned control variables. This aims to simplify the model structure and more directly assess the impact of sustainable livelihood capital factors on women’s decision-making regarding participation in productive and operational activities. According to the estimation results in Table 2, based on the Bayesian Information Criterion (BIC) [39], Model M1 has a lower BIC value and a more refined variable structure, indicating that it achieves a better trade-off between model fit and parsimony. Therefore, this paper ultimately selects Model M1 as the main model to explain the key factors influencing the decision of the female labor force to participate in productive and operational activities.

Table 2. Results of baseline regression.

Variables		M1: Logit Model Without Controls		M2: Logit Model
		Coefficient	z-Value	Coefficient	z-Value
	age_56plus			−1.283 **	−2.232
Control variables	hh_large			−1.400 ***	−2.685
	marital
	Married			−0.54	−0.995
	Divorced			−1.844 *	−1.760
	Widowed			−0.600	−0.488
	near_fish_port			−1.394 **	−2.529
Social capital	gender_div
	gender equal	1.039 *	1.797	1.357 **	1.992
	male important	0.592	1.334	0.317	0.672
	Minimal diff	0.555	0.492	0.263	0.177
	skill_income
	Prob increase	0.469	0.873	1.202 *	1.899
	Uncertain	1.137 *	1.848	1.564 **	2.188
	No increase	0.813	0.837	1.935 *	1.766
Human capital	edu_high	1.004 ***	2.58	0.835 *	1.764
	female_labor	0.298 *	1.684	0.688 ***	2.699
	income_stable	1.808 ***	4.116	2.116 ***	4.114
	training_effect	1.301 ***	2.669	1.516 ***	2.691
Physical capital	boat_asset	−1.492 ***	−3.205	−1.357 ***	−2.942
	waterway_infra	−2.933 ***	−3.461	−2.039 **	−2.210
Financial capital	credit_access	3.064 ***	3.731	3.668 ***	3.453
	income_share
	25%–50%	−1.246 **	−2.164	−1.002	−1.636
	50%–75%	−0.918 *	−1.695	−1.640 ***	−2.740
	Above 75%	−3.221 ***	−5.807	−3.663 ***	−5.752
Natural capital	over_logging	1.479 ***	2.993	2.038 ***	3.939
	Constant	−2.218	−1.514	−3.642 **	−1.978
N		250		250
AIC		244.6		239.5
BIC		308		324
Pseudo R2		0.386		0.436

Notes: ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Robust standard errors are reported in parentheses. The same convention applies to subsequent tables.

From a social capital perspective, an equal household division of labor significantly increases the probability of women participating in productive activities compared to those in low-income work (significant at the 10% level). This positive effect persists even for female laborers uncertain about skill-to-income effectiveness (10% significance level). External constraints, such as labor market environments and family responsibilities, moderate the effect, validating Hypothesis H1.

Regarding physical capital, a paradox emerges: owning boats and machinery for aquaculture or fishing, along with improvements in waterway transportation infrastructure, reduces the probability of female labor participation in productive and operational activities. The reason may be that these activities are mainly undertaken by men, particularly offshore fishing or operating large-scale machinery. Although improvements in waterway transportation may have contributed to the development of related sectors, such as aquatic product transport and mangrove-based tourism, the employment opportunities created appear to show clear gender disparities. High-value roles requiring physical strength or technical skills (e.g., loading/unloading, boat piloting, tourist boat services) are predominantly occupied by men. Female participation is largely limited to auxiliary, seasonal, and low-value activities. These opportunities are often mediated by local communities or external businesses, further constraining women’s economic empowerment. Limited involvement in key decision-making, from organizational arrangements to the distribution of economic benefits, also reduces incentives for women to engage in independent productive activities. These factors collectively explain why improvements in waterway transportation conditions have a significant negative impact on female labor participation, validating Hypothesis H2.

In terms of human capital, women with a high school education or above have a significantly higher probability of participating in productive and operational activities. This result indicates that education helps them enhance their ability to access market information, strengthen operational skills, and improve their capacity to adapt to business, tourism, and service-related activities. Conversely, women with lower education levels often marry earlier and primarily bear household responsibilities, which limits their participation to some extent. Similarly, the number of female laborers aged 15 and above in the household has a significant positive impact, indicating that more abundant female labor resources allow flexible division and arrangement of housework, thereby creating more favorable conditions for women to participate in productive and operational activities. Notably, the perception that mangrove-related work ensures a stable livelihood is positively associated with engagement, suggesting that the perception of income stability may drive participation. Related activities are mainly concentrated in aquaculture, aquaculture logistics, and processing, often near residential areas, making them more suitable for married women. Ensuring livelihood stability also helps them balance economic labor and family responsibilities. Furthermore, the perception of the practicality of skills enhancement training has a significant positive impact. These results validate Hypothesis H3.

In comparison, financial capital plays a significant role in promoting the participation of the female labor force in productive and operational activities. Access to credit for mangrove-related economic activities has a significant positive impact, enabling women to invest in tourism node construction, small-scale entrepreneurial projects, production expansion, processing facilities, and commercial/service businesses. However, when household income heavily depends on mangrove resources, with forestry-related income accounting for approximately 75% of total household income, this factor shows a negative association, suggesting that reliance on natural resource-based income may be associated with less livelihood diversification and a relatively lower likelihood of participation in non-agricultural productive and operational activities. These results validate Hypothesis H4.

Finally, regarding natural capital, perceptions of ecological degradation and overexploitation of mangrove resources are positively correlated with female participation, suggesting a shift toward more sustainable livelihood strategies, thus validating Hypothesis H5. This finding implies that environmental perceptions not only influence individual attitudes toward resource protection but also play an important role in shaping the long-term livelihood choices and development paths of the female labor force in mangrove areas.

4.2. Robustness Checks

To ensure the robustness and reliability of the baseline model, this study conducted two main robustness tests. First, we used a binary Probit model instead of a Logit model. Second, we apply the Entropy Weight Method to determine the weight of each indicator objectively based on the data itself, thereby minimizing subjectivity in assigning weights and partly solving the problem of measurement error. These models examine the impact of five types of livelihood capital (natural, physical, financial, human, and social) on the probability of female labor participation in productive and operational activities in mangrove areas, while controlling for factors such as age, household size, marital status, and geographical location. According to the results in Table 3, the Probit model (M4) produces results that are completely consistent with those of the Logit model (M2). All key variables maintain the same positive or negative direction of their coefficients, and the statistical significance levels are generally similar across the two models. Although the significance levels of some explanatory variables slightly decrease in the Probit model, the direction of the estimated coefficients remains consistent, and the model fit evaluation indicators (pseudo-R² and BIC) show almost no change. This indicates that the estimation results are not sensitive to the choice of the error term distribution, thereby confirming the robustness and reliability of the Logit regression results. The EWM results are presented in Table 4, social capital had the highest weight (0.515), followed by human capital (0.159) and natural capital (0.149); physical capital and financial capital had weights of 0.121 and 0.056, respectively. Entropy values ranged from 0.3766 to 0.9903, reflecting the different levels of variability among the indicators. In this study, the minimal_division (entropy = 0.3766) and over_logging (entropy = 0.6726) indicators showed the highest dispersion, and therefore were assigned higher weights by the EWM method.

Table 3. Results of robustness checks.

Variables		M3: Probit Without Controls		M4: Probit Full Model
		Coefficient	z-Value	Coefficient	z-Value
	age_56plus			−0.768 **	−2.412
Control variables	hh_large			−0.770 ***	−2.699
	marital
	Married			−0.314	−0.984
	Divorced			−1.126 **	−1.966
	Widowed			−0.384	−0.573
	near_fish_port			−0.796 ***	−2.647
Social capital	gender_div
	gender equal	0.545	1.63	0.721 *	1.926
	male important	0.29	1.193	0.123	0.489
	Minimal diff	0.276	0.458	0.181	0.245
	skill_income
	Prob increase	0.148	0.473	0.545	1.457
	Uncertain	0.527	1.493	0.739 *	1.855
	No increase	0.261	0.486	0.883	1.402
Human capital	edu_high	0.537 **	2.476	0.444 *	1.754
	female_labor	0.162	1.584	0.366 ***	2.601
	income_stable	1.001 ***	4.126	1.150 ***	4.117
	training_effect	0.662 **	2.41	0.745 **	2.353
Physical capital	boat_asset	−0.824 ***	−3.307	−0.762 ***	−3.037
	waterway_infra	−1.698 ***	−3.518	−1.156 **	−2.254
Financial capital	credit_access	1.759 ***	4.016	2.092 ***	3.659
	income_share
	25%–50%	−0.720 **	−2.223	−0.592 *	−1.751
	50%–75%	−0.485	−1.597	−0.911 ***	−2.726
	Above 75%	−1.784 ***	−5.836	−2.009 ***	−5.658
Natural capital	over_logging	0.839 ***	3.045	1.141 ***	3.813
	Constant	−1.029	−1.253	−1.749 *	−1.699
N		250		250
AIC		246		241
BIC		309.4		325.5
Pseudo R2		0.381		0.432

Notes: ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Robust standard errors are reported in parentheses.

Table 4. Weights of indicators by capital type within the sustainable livelihood framework.

Capital Type	Weight	Weight	Entropy
Social capital	skill_income	0.515	0.9746
	gender_div
	gender equal		0.6439
	male important		0.8745
	Minimal division		0.3766
Human capital	edu_high	0.159	0.7959
	female_labor		0.9766
	income_stable		0.9415
	training_effective		0.9375
Physical capital	boat_asset	0.121	0.7475
	waterway_infra		0.9865
Financial capital	credit_access	0.056	0.9903
	income_share		0.8869
Natural capital	over_logging	0.149	0.6726

Based on these weights, the study performs a robustness check by re-estimating the regression model with four alternative specifications. Specifically, the main model is estimated with four specifications: Logit with control variables, Logit without controls, Probit with controls, and Probit without controls. The capital variables in these models are composite scores (Human Capital Score, Physical Capital Score, Financial Capital Score, Natural Capital Score, and Social Capital Score), each aggregated from multiple underlying indicators as described in Section 3.2. The results in Table 5 demonstrate a high degree of consistency across all four specifications. Human Capital (HC_score) and Financial Capital (FC_score) are statistically significant at the 1% level with the same sign (positive for HC_score, negative for FC_score) in every model. Natural capital (NC_score) is statistically significant at the 5%–10% level across all models. Physical capital (PC_score) is statistically significant only at the 10% level in models without control variables (Logit Basic and Probit Basic). Social Capital (SC_score) remains positive but not statistically significant in all models. Furthermore, the control variables, particularly age_56plus, divorced marital status, and near_fish_port–show negative and statistically significant effects in both Logit and Probit full models, further confirming the stability of our estimates. The BIC values are slightly lower in the basic models (without controls), suggesting a trade-off between model fit and parsimony; however, the direction and significance of the main capital variables do not change when controls are added, indicating that our core findings are not driven by demographic or geographic factors.

Table 5. Results of robustness check based on EWM.

Variables		Logit Basic		Logit Full		Probit Basic		Probit Full
		Coefficient	z-Value	Coefficient	z-Value	Coefficient	z-Value	Coefficient	z-Value
Social Capital	SC_score	2.015	0.8	3.015	1.1	1.018	0.71	1.538	1.01
Human Capital	HC_score	14.176 ***	4.24	13.828 ***	3.49	8.505 ***	4.39	8.331 ***	3.66
Physical Capital	PC_score	−5.591 *	−1.83	−4.736	−1.43	−3.284 *	−1.84	−2.703	−1.39
Financial Capital	FC_score	−49.944 ***	−6.59	−53.831 ***	−6.52	−29.858 ***	−6.98	−31.696 ***	−6.93
Natural Capital	NC_score	5.777 *	1.76	7.598 **	2.04	3.627 *	1.87	4.510 **	2.09
Control variables	age_56plus			−0.797 *	−1.67			−0.452	−1.64
	hh_large			−0.418	−1.03			−0.271	−1.14
	marital
	Married			−0.193	−0.34			−0.149	−0.43
	Divorced			−1.620 **	−2.03			−0.988 **	−2.08
	Widowed			−0.33	−0.32			−0.225	−0.38
	near_fish _port			−1.166 **	−2.26			−0.683 **	−2.22
	_cons	0.780 **	2.13	1.452 **	2.03	0.476 **	−2.19	0.889 **	2.1
	N	250		250		250		250
	AIC	267.104		264.626		266.206		263.936
	BIC	288.233		306.883		287.334		306.194
	Pseudo R2	0.249		0.291		0.251		0.293

Notes: ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Robust standard errors are reported in parentheses.

In summary, both robustness checks indicate that the main conclusions of the study do not change significantly when altering the model specification (Logit vs. Probit) or the indicator weighting method (EWM). A comparison between the models using disaggregated indicators (Table 2 and Table 3) and those using composite scores derived from EWM (Table 5) reveals that the direction of effects for human, natural, physical, and social capital is consistent across all eight models. Regarding financial capital, however, there is a discrepancy between the two measurement approaches. Overall, the findings of the study are further reinforced in terms of reliability.

4.3. Heterogeneity Analysis

(1)

Regression results of sustainable livelihood capital on productive and operational activities of the female labor force across different age groups

Based on the results in Table 6 and Table 7, the Logit regression results for female laborers under 56 years old (age_56plus = 0) show that the model has good explanatory power (Pseudo R² = 0.379), reflecting the complex interactions among different types of livelihood capital in shaping participation decisions in productive and operational activities. In contrast, for female laborers aged 56 and above, the limited sample size and the frequent occurrence of perfect prediction in small-sample models make regression analysis less reliable. Therefore, this study primarily uses descriptive statistics and cross-tabulation analysis to highlight trends in livelihood strategies as age increases.

Table 6. Heterogeneity analysis based on individual characteristics of the female labor force.

Variables		M5: Age < 56		M6: Married		M7: Household Size		M8: Non-Fishing-Port Area
		Coef.	z-Value	Coef.	z-Value	Coef.	z-Value	Coef.	z-Value
Social capital	gender_div
	gender equal	0.904	1.265	0.79	1.187	1.465 *	1.951	0.792	1.254
	male important	0.175	0.36	0.31	0.626	0.786	1.501	0.356	0.79
	Minimal diff	0.376	0.312	−1.104	−1.204	0.851	0.73	0.123	0.104
	skill_income
	Prob increase	0.715	1.097	0.895	1.386	0.553	0.805	0.643	1.059
	Uncertain	1.032	1.516	1.638 **	2.375	0.863	1.136	1.354 *	1.921
	No increase	0.753	0.542	1.076	0.839	1.18	0.925	0.845	0.792
Human capital	edu_high	0.884 **	2.034	1.077 **	2.292	0.859 *	1.793	0.675	1.58
	female_labor	0.296	1.255	0.353	1.284	0.622 **	2.06	0.228	1.313
	income_stable	1.745 ***	3.683	1.386 ***	2.666	1.757 ***	3.444	1.886 ***	3.858
	training_effect	1.227 **	2.237	1.295 **	2.164	1.278 **	2.294	1.401 ***	2.626
Physical capital	boat_asset	−1.468 ***	−2.905	−1.814 ***	−3.392	−0.552	−1.031	−1.180 **	−2.360
	waterway_infra	−2.888 ***	−3.089	−2.700 ***	−3.081	−2.575 ***	−3.218	−2.806 ***	−3.342
Financial capital	credit_access	3.372 ***	3.443	2.498 ***	3.489	3.129 ***	3.251	3.406 ***	3.925
	income_share
	25%–50%	−0.755	−1.044	−1.426 **	−2.185	−0.19	−0.302	−1.215 *	−1.895
	50%–75%	−0.908	−1.445	−1.308 **	−2.136	−0.494	−0.764	−1.222 **	−2.050
	Above75%	−3.115 ***	−4.987	−3.547 ***	−5.121	−3.218 ***	−4.752	−3.208 ***	−5.155
Natural capital	over_logging	2.332 ***	3.745	1.468 ***	2.661	2.433 ***	3.964	2.228 ***	3.402
	Constant	−2.327	−1.413	−1.509	−1.015	−3.541 *	−1.950	−2.537	−1.630
AIC		207.4		199.4		188.8		217.7
BIC		267.7		258.5		248		278.6
Pseudo R2		0.379		0.387		0.423		0.382

Notes: ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively.

Table 7. Heterogeneity analysis based on individual characteristics of the female labor force A comparative analysis of female labor participation in production and business activities across different age cohorts.

Variables		Level	age_56plus = 0 (N = 210)	age_56plus = 1 (N = 40)	p-Value F Statistic (df)
Social capital	gender_div				0.037 8.49 (3)
		female led	68 (32.4%)	14 (35.0%)
		gender equal	24 (11.4%)	11 (27.5%)
		male important	111 (52.9%)	14 (35.0%)
		Minimal division	7 (3.3%)	1 (2.5%)
	skill_income				<0.001 30.67 (3)
		Def increase	44 (21.0%)	5 (12.5%)
		Prob increase	111 (52.9%)	10 (25.0%)
		Uncertain	47 (22.4%)	15 (37.5%)
		No increase	8 (3.8%)	10 (25.0%)
Human capital	edu_high		80 (38.1%)	1 (2.5%)	<0.001 19.44 (1)
	female_labor, mean (SD)		1.65714 (0.845235)	2.375 (1.33373)	<0.001 19.64 (1, 248)
	income_stable		154 (73.3%)	27 (67.5%)	0.45 0.57 (1)
	training_effect		160 (76.2%)	17 (42.5%)	<0.001 18.45 (1)
Physical capital	boat_asset		47 (22.4%)	15 (37.5%)	0.042 4.12 (1)
	waterway_infra		193 (91.9%)	39 (97.5%)	0.21 1.57 (1)
Financial capital	credit_access		201 (95.7%)	36 (90.0%)	0.14 2.23 (1)
	income_share				0.026 9.26 (3)
		Below 25%	97 (46.2%)	11 (27.5%)
		25%–50%	18 (8.6%)	7 (17.5%)
		50%–75%	35 (16.7%)	4 (10.0%)
		Above 75%	60 (28.6%)	18 (45.0%)
Natural capital	over_logging		34 (16.2%)	7 (17.5%)	0.84 0.04 (1)

Regarding human and financial capital, these two types correlate positively with the participation of women under 56. Educational attainment demonstrates a positive association with the capacity for participation, while livelihood stability also appears to exhibit a positive correlation. This indicates that women with higher education and stronger confidence in the sustainability of the mangrove ecosystem are more likely to engage in entrepreneurial activities or expand service-oriented businesses. Access to credit also exhibits a positive association, suggesting that loan capital may serve as an important mechanism to help women navigate initial investment constraints and facilitate entry into the market. For women aged 56 and above, although regression analysis was not conducted, descriptive statistics indicate that only 2.5% have higher education, and most productive activities rely on practical experience rather than modern business models. Regarding physical and natural capital, reverse effects are observed. Ownership of productive tools such as fishing boats and high dependence on mangrove income appear to be negatively associated with the probability of participation. This suggests that when households already rely on traditional fishing tools and stable fishing income, their motivation to explore more complex service-oriented business activities decreases. However, the natural capital variable (over_logging) suggests that when overexploitation occurs, women under 56 may turn to alternative productive and operational activities as a possible adaptive response to resource depletion. In terms of social capital, households with a more gender-equal division of labor tend to have higher participation probabilities for women compared with households where women are primarily engaged in low-income work. However, this relationship is not statistically significant in the Logit regression. When men dominate household labor, women’s participation probability decreases significantly, showing that households with relatively balanced social capital are more conducive to women’s engagement. Cross-tabulation results suggest that social capital influences participation trends through household labor division, though it is not a direct determinant when other livelihood capitals are controlled.

Overall, women under 56 mainly rely on education, human capital, and financial capital to participate in productive and operational activities, while also benefiting from skills training. In contrast, women aged 56 and above, although experienced, are constrained by limited education and training opportunities, and the small sample size restricts quantitative analysis. Therefore, for older women, descriptive statistics and cross-tab.

(2)

Regression results of sustainable livelihood capital and productive and operational activities of the female labor force by marital status

Based on Table 6 and Table 8, the Logit regression results indicate that married women’s participation in productive and operational activities is significantly influenced by various types of livelihood capital. However, due to differences in sample sizes between the married group and other groups (single, divorced, and widowed women), the study also employs descriptive statistics and chi-square tests to capture behavioral trends among minority groups and to reduce biases caused by perfect prediction in small samples.

Table 8. Differences in productive and operational activities of the female labor force by marital status.

	Variables	Level	Single (N = 25)	Married (N = 197)	Divorced (N = 18)	Widowed (N = 10)	p-Value F Statistic (df)
Social capital	gender_div						0.46 8.71 (9)
		female led	6 (24.0%)	65 (33.0%)	6 (33.3%)	5 (50.0%)
		gender equal	3 (12.0%)	30 (15.2%)	1 (5.6%)	1 (10.0%)
		male important	16 (64.0%)	96 (48.7%)	9 (50.0%)	4 (40.0%)
		Minimal division	0 (0.0%)	6 (3.0%)	2 (11.1%)	0 (0.0%)
	skill_income						0.071 15.83 (9)
		Def increase	7 (28.0%)	40 (20.3%)	1 (5.6%)	1 (10.0%)
		Prob increase	15 (60.0%)	90 (45.7%)	13 (72.2%)	3 (30.0%)
		Uncertain	2 (8.0%)	54 (27.4%)	2 (11.1%)	4 (40.0%)
		No increase	1 (4.0%)	13 (6.6%)	2 (11.1%)	2 (20.0%)
Human capital	edu_high	-	20 (80.0%)	53 (26.9%)	8 (44.4%)	0 (0.0%)	<0.001 34.56 (3)
	female_labor	Mean (SD)	2.16 (1.02)	1.69 (0.85)	1.44 (0.92)	2.9 (1.97)	<0.001 7.42 (3, 246)
	income_stable	-	20 (80.0%)	148 (75.1%)	9 (50.0%)	4 (40.0%)	0.011 11.23 (3)
	training_effect	-	16 (64.0%)	147 (74.6%)	12 (66.7%)	2 (20.0%)	0.002 14.58 (3)
Physical capital	boat_asset	-	5 (20.0%)	54 (27.4%)	0 (0.0%)	3 (30.0%)	0.068 7.11 (3)
	waterway_infra	-	24 (96.0%)	182 (92.4%)	16 (88.9%)	10 (100.0%)	0.65 1.62 (3)
Financial capital	credit_access	-	25 (100.0%)	185 (93.9%)	18 (100.0%)	9 (90.0%)	0.37 3.14 (3)
	income_share						0.46 8.72 (9)
		Below 25%	11 (44.0%)	83 (42.1%)	11 (61.1%)	3 (30.0%)
		25%–50%	0 (0.0%)	22 (11.2%)	2 (11.1%)	1 (10.0%)
		50%–75%	4 (16.0%)	33 (16.8%)	0 (0.0%)	2 (20.0%)
		Above 75%	10 (40.0%)	59 (29.9%)	5 (27.8%)	4 (40.0%)
Natural capital	over_logging	-	0 (0.0%)	37 (18.8%)	3 (16.7%)	1 (10.0%)	0.11 6.02 (3)

In terms of human and social capital, education level (edu_high) and confidence in income stability (income_stable) are key factors driving married women’s participation in productive and operational activities. In contrast, cross-tabulations show that the proportion of single women with higher education is substantially higher (80%), but the number of female laborers in their households is lower than in the widowed women group, suggesting that single women have a knowledge advantage for modern production and operation but lack internal support networks. Married women also rated training effectiveness higher (training_effect, p = 0.002; 74.6%) compared to widowed women (20%), suggesting that married women may be more inclined to upgrade their skills for managing productive and operational activities, whereas single or divorced women appear more likely to maintain their current livelihoods. Regarding physical and financial capital, a clear resource dependence paradox emerges. Among married female laborers, regression analysis identifies two factors that significantly reduce participation probability: the possession of productive assets such as boats (boat_asset) and critical reliance on mangrove income (income_share > 75%). This reflects path dependency, where past investments and livelihood choices constrain the ability to transition to new productive and operational activities, even when better economic opportunities exist. In contrast, divorced women and widowers rarely own boats but most have access to credit (credit_access nearly 100%), showing higher motivation in using financial resources for small business operations or service enterprises. In terms of natural capital, over-logging (over_logging) is positively associated with participation. This finding could be interpreted as reflecting adaptive strategies under resource depletion. Married and divorced women are more sensitive to environmental degradation (18.8% and 16.7%, respectively) compared to single women (0%), likely due to household responsibilities, prompting them to actively seek non-agricultural livelihood options when mangrove resources decline.

Taken together, married women appear to draw upon human and financial capital in adapting their livelihood activities under conditions of declining natural resources, whereas single and divorced women show some tendency toward livelihood adjustment through education and credit, even with limited physical assets. These results highlight that livelihood support policies should account for marital status heterogeneity, providing skills training for married women and social networks along with financial support for single women, thereby enhancing economic participation across different groups. However, it should be noted that due to sample size limitations, these findings need further validation with larger samples before definitive conclusions can be drawn.

(3)

Regression results of sustainable livelihood capital and productive and operational activities of the female labor force by household size

Based on Table 6 and Table 9, given the significant differences in sample sizes and the occurrence of perfect prediction in some small-sample subgroups (household size ≥ 6, hh_large = 1), this study primarily conducts regression analysis on households with fewer than six members. Descriptive statistics and cross-tabulation analysis are applied to larger households to identify behavioral trends while avoiding estimation bias due to small sample sizes.

Table 9. Differences in production and business activities of female labor force by household size.

	Variables	Level	hh_large = 0 (<6 People) (N = 198)	hh_large = 1 (≥6 People) (N = 52)	p-Value F Statistic (df)
Social capital	gender_div				0.51 2.31 (3)
		female led	65 (32.8%)	17 (32.7%)
		gender equal	28 (14.1%)	7 (13.5%)
		male important	97 (49.0%)	28 (53.8%)
		Minimal division	8 (4.0%)	0 (0.0%)
	skill_income				0.019 9.95 (3)
		Def increase	42 (21.2%)	7 (13.5%)
		Prob increase	90 (45.5%)	31 (59.6%)
		Uncertain	55 (27.8%)	7 (13.5%)
		No increase	11 (5.6%)	7 (13.5%)
Human capital	edu_high	-	64 (32.3%)	17 (32.7%)	0.96 0.00 (1)
	female_labor	Mean (SD)	1.57071 (0.782129)	2.53846 (1.22813)	<0.001 48.45 (1, 248)
	income_stable	-	135 (68.2%)	46 (88.5%)	0.004 8.48 (1)
	training_effect	-	137 (69.2%)	40 (76.9%)	0.28 1.19 (1)
Physical capital	boat_asset	-	42 (21.2%)	20 (38.5%)	0.01 6.57 (1)
	waterway_infra	-	180 (90.9%)	52 (100.0%)	0.024 5.09 (1)
Financial capital	credit_access	-	187 (94.4%)	50 (96.2%)	0.62 0.24 (1)
	income_share				0.38 3.07 (1)
		Below 25%	91 (46.0%)	17 (32.7%)
		25%–50%	19 (9.6%)	6 (11.5%)
		50%–75%	30 (15.2%)	9 (17.3%)
		Above 75%	58 (29.3%)	20 (38.5%)
Natural capital	over_logging	-	36 (18.2%)	5 (9.6%)	0.14 2.20 (1)

Human capital and financial capital are among the factors positively associated with female labor force participation. The regression results indicate that in larger households (hh_large = 1), human capital has a significant promoting effect on women’s engagement in productive and operational activities. The number of female laborers in the household (female_labor) shows a positive association with participation. This could be interpreted as reflecting that a greater local labor supply helps share household responsibilities, potentially creating time and conditions for women to engage in production and business. Confidence in livelihood stability (income_stable) and recognition of training effectiveness (training_effect) further strengthen women’s autonomous decision-making. Most importantly, access to credit (credit_access) has the strongest positive effect, highlighting that financial resources are a prerequisite for the female labor force to overcome initial cost barriers when entering non-agricultural industries. In contrast, physical and natural capital appear to be negatively associated with livelihood diversification. High dependence on forest income (income_share > 75%) and limitations in waterway infrastructure (waterway_infra) may act as underlying constraints, reinforcing continued engagement in traditional livelihood paths and potentially limiting the scope for women to shift toward more diversified productive and operational activities. However, over-logging behavior (over_logging) shows a positive correlation with participation, reflecting a passive adaptive strategy: as natural resources decline and traditional livelihoods become harder to sustain, women are compelled to explore new business models to maintain household livelihoods. Descriptive statistics show that large households (hh_large = 1) have higher rates of boat ownership and perceive greater livelihood stability compared to smaller households. Nevertheless, for older female laborers, cross-tabulation analysis reveals that lower education levels and reduced confidence in effectiveness correlate negatively with engagement in more complex production and operational activities. Even in large households of six or more members, this group primarily participates in auxiliary labor or primary resource extraction rather than activities requiring higher technical skills and organizational capacity.

In summary, female labor participation in productive and operational activities results from the interaction between the advantages of human and financial capital and the pressures of natural resource degradation. Policy design should prioritize credit access and skills training to fully utilize the labor potential of women in larger households, while also implementing differentiated adaptive support measures for older female laborers to alleviate structural constraints and facilitate the transition to diversified livelihoods.

(4)

Regression results of sustainable livelihood capital on production and operational activities of female labor in different geographic locations

Based on Table 5 and Table 9, for female laborers in areas far from fishing ports, the Logit regression results show a high overall model fit, clearly identifying the key factors influencing women’s participation in production and business activities. In contrast, for women living near fishing ports, wharves, or tidal flats (near_fish_port = 1), regression estimates are less stable due to the small sample size. Therefore, this study uses the regression results from areas far from fishing ports as the primary analytical basis, while supplementing with descriptive statistics and cross-tabulations to examine the livelihood characteristics and behavioral trends of women near ports, ensuring the scientific rigor and objectivity of statistical inference.

In areas far from fishing ports, human capital is the main driving force promoting women’s participation. Confidence in income stability (income_stable) and perceived training effectiveness (training_effect) have positive impacts on engagement. Skills training enhances capabilities, facilitates entry into non-agricultural value chains, and supports livelihood transformation. In contrast, cross-tabulation analysis shows that in areas near ports, 40.6% of women believe skills will definitely increase income (Def_increase), compared to 16.5% in areas far from ports. This indicates that in market node areas such as fishing ports, the economic value of human capital and specialized skills is more readily recognized and rewarded by market mechanisms, thereby strengthening women’s intrinsic motivation to engage in production and business activities. A paradox emerges regarding physical and financial capital. In distant areas, owning fishing boats (boat_asset) and having waterway infrastructure (waterway_infra) is negatively associated with participation probability, which may reflect a traditional asset trap where physical capital becomes locked into resource-intensive activities and extensive production pathways, limiting livelihood diversification. In contrast, financial capital, particularly access to credit (credit_access), demonstrates a leverage effect, enhancing women’s ability to reduce dependence on forest resources and engage in non-agricultural activities. Near fishing ports, 37.5% of women own boats and nearly 100% have access to credit, suggesting that fishing ports serve not only as commodity circulation centers but also as key nodes for capital flow, effectively lowering financial barriers for women to participate in market-oriented production and business activities. Natural capital also plays a role: over-logging shows a positive correlation with participation, which may reflect a crisis-driven adaptation strategy. In distant areas, women are compelled to pursue commercial or service-oriented activities as mangrove resources decline. Near ports, although 18.8% recognize resource degradation, more diverse and secure market options reduce the risks involved in livelihood transformation.

Overall, geographic proximity to fishing ports acts as an amplifier, significantly strengthening the positive effects of human and financial capital on women’s participation in production and business activities. Conversely, women in areas far from ports face dual constraints from limited infrastructure and resource dependence. These findings highlight that information accessibility and financial availability in port areas are key institutional conditions for promoting more active and sustainable participation of women in production and business (Table 10).

Table 10. Differences in production and business activities of female labor force by geographic location.

	Variables	Level	near_fish_port = 0 Not Near Fishing Port (N = 218)	near_fish_port = 1 Near Fishing Port (N = 32)	p-Value F Statistic (df)
Social capital	gender_div				0.2 4.66 (3)
		female led	73 (33.5%)	9 (28.1%)
		gender equal	27 (12.4%)	8 (25.0%)
		male important	110 (50.5%)	15 (46.9%)
		Minimal division	8 (3.7%)	0 (0.0%)
	skill_income				0.008 11.95 (3)
		Def increase	36 (16.5%)	13 (40.6%)
		Prob increase	109 (50.0%)	12 (37.5%)
		Uncertain	55 (25.2%)	7 (21.9%)
		No increase	18 (8.3%)	0 (0.0%)
Human capital	edu_high	-	66 (30.3%)	15 (46.9%)	0.061 3.51 (1)
	female_labor	Mean (SD)	1.74 (0.98)	1.97 (0.93)	0.22 1.50 (1, 248)
	income_stable	-	158 (72.5%)	23 (71.9%)	0.94 0.01 (1)
	training_effect	-	155 (71.1%)	22 (68.8%)	0.78 0.07 (1)
Physical capital	boat_asset	-	50 (22.9%)	12 (37.5%)	0.075 3.17 (1)
	waterway_infra	-	200 (91.7%)	32 (100.0%)	0.092 2.85 (1)
Financial capital	credit_access	-	205 (94.0%)	32 (100.0%)	0.16 2.01 (1)
	income_share				0.075 6.91 (3)
		Below 25%	91 (41.7%)	17 (53.1%)
		25%–50%	21 (9.6%)	4 (12.5%)
		50%–75%	39 (17.9%)	0 (0.0%)
		Above 75%	67 (30.7%)	11 (34.4%)
Natural capital	over_logging	-	35 (16.1%)	6 (18.8%)	0.7 0.15 (1)

5. Discussion

The findings of this study contribute to reinforcing and extending existing understandings of the relationship between livelihood capital and women’s economic participation in mangrove-dependent areas. In the first respect, the results indicate that financial and human capital play an important role in facilitating women’s engagement in income-generating activities. This observation is broadly consistent with perspectives articulated in prior work emphasizing the importance of credit access and educational attainment for the economic empowerment of women from disadvantaged groups in rural and coastal settings [41,42]. In this regard, the present study offers additional empirical support for these assertions within the specific context of the mangrove areas of Ca Mau. In the second respect, the study also tentatively identifies certain notable divergences from previous analyses. Specifically, while reaffirming the salience of financial and human capital, the results suggest that excessive reliance on mangrove-based natural resources—particularly when such resources constitute a predominant share of household income—appears to be associated with restricted movement of women away from traditional livelihood activities. Although prior research has acknowledged the vulnerability of resource-dependent livelihoods [43,44], this finding helps to further elucidate the mechanism of resource “lock-in” within the distinctive socio-ecological context of mangrove ecosystems and points to a more intricate interplay among livelihood capitals. Moreover, the results hint that the role of physical capital may not be uniformly enabling as observed in some other settings; rather, in this instance, it seems to be considerably shaped by prevailing gendered divisions of labor. On balance, while the majority of the findings lend support to the sustainable livelihoods framework and related empirical inquiries, this study also highlights several context-specific nuances, thereby modestly enriching the understanding of the factors shaping women’s economic participation in Vietnam’s coastal communities.

This study examines how livelihood capital shapes women’s labor participation in mangrove-dependent households in Cà Mau province, Vietnam. Using survey data and regression analysis, four main findings emerge. First, financial and human capital are core drivers of women’s economic participation. Access to credit helps women overcome initial capital constraints, while higher education and vocational training improve their market adaptability and operational autonomy, particularly among women under 56 years old. Second, a livelihood paradox exists. Heavy reliance on mangrove resources (over 75% of income) locks women into traditional, male-dominated extraction activities. Physical capital, rather than enabling women’s economic roles, reinforces gender-based labor division. Third, natural capital degradation triggers a crisis-driven transition. As mangrove resources decline, women are compelled to shift toward non-farm activities (small trade, services) to maintain household livelihood security, thereby promoting livelihood diversification. Fourth, women’s participation is heterogeneous. Larger household size constrains participation, but a higher number of female workers in the household reallocates domestic labor and creates positive effects. Geographic proximity to fishing ports amplifies the facilitating effects of livelihood capital. Married women prioritize livelihood stability and training benefits, while single or divorced women—despite lacking physical capital—achieve livelihood transitions through better credit access. These findings provide empirical evidence for integrating gender, livelihood diversification, and ecological conservation policies in mangrove-dependent coastal regions. However, given the single-province sample, future research should replicate this analysis in other contexts before drawing broader generalizations

Although this study has achieved several important findings, certain limitations need to be acknowledged. These limitations reflect the inherent constraints of the current research design and simultaneously open up directions for future research. First, a limitation relates to the measurement and classification of livelihood capital types. Due to sample size constraints and survey design, natural capital is represented by only a single proxy variable (over-exploitation behavior), which fails to fully capture other important dimensions such as mangrove area, water quality, or saltwater intrusion. Furthermore, the boundaries between social capital, human capital, and demographic characteristics within the DFID sustainable livelihoods framework overlap to some extent, which may affect the ability to clearly disentangle the distinct effects of each capital type. Therefore, future research should expand the set of indicators and apply exploratory factor analysis (EFA) or confirmatory factor analysis (CFA) to clarify the structure and boundaries of each livelihood capital type across different socio-cultural contexts.

Second, a limitation concerns causal inference and potential endogeneity. Although the survey design collected independent variables reflecting the situation in the previous year and the dependent variable reflecting the current year’s situation, which may help mitigate concerns about reverse causality, endogeneity might not have been fully addressed. Moreover, given the cross-sectional design, the results may primarily reflect correlations rather than causal relationships–an inherent limitation of this approach. Therefore, future research could consider constructing more suitable instrumental variables to further investigate causal relationships.

Third, a limitation relates to the sample and external validity. Several indicators exhibit relatively high mean values (e.g., credit_access: 0.948; waterway_infra: 0.928), resulting in limited data variation. Furthermore, the variable age_56plus reflects the legal retirement age (56 years) rather than individuals’ actual health status. Additionally, the study may be subject to a potential risk of social desirability bias, particularly for questions concerning gender roles–a type of bias that could affect the reliability of the reported correlations. Furthermore, the study was conducted in a single province (Cà Mau) using self-reported data, so the findings may not be generalizable to other ecological, cultural, or economic contexts. Therefore, this study recommends that future research should expand both sample size and geographical scope. Specifically, the analysis should be replicated across different regions of Vietnam (as well as in other countries with similar mangrove ecosystems), while also employing longitudinal or quasi-experimental designs to test the external validity of the conclusions.

6. Conclusions

This study investigates how different types of livelihood capital affect women’s labor participation in mangrove-dependent areas in Cà Mau Province, Vietnam, based on survey data from 250 households and quantitative analyses. The effects of different capital types vary in direction and magnitude, reflecting complex structural patterns in women’s economic engagement. Synthesizing the empirical results, the following four main conclusions can be drawn. First, financial capital and human capital appear to be important drivers promoting women’s economic participation. Access to credit and education may significantly support women’s ability to engage in higher-value production and business activities, particularly among women under 56 years of age. Second, there is evidence suggesting the existence of a resource lock-in effect. Heavy reliance on natural resources may have limited women’s ability to shift toward diversified livelihoods. Meanwhile, physical capital does not appear to have been transformed into productive assets under women’s control; instead, it tends to reinforce the gendered division of labor, thereby potentially creating a structural barrier to women’s participation in modern business activities. Third, the degradation of natural capital may have generated a pressure-driven transition process. When traditional resource-based livelihoods become less sustainable, women tend to shift toward small-scale trading and service activities, through which they may contribute to the diversification of household income sources. Fourth, women’s labor participation is simultaneously influenced by household structure, marital status, and geographical location. Larger households tend to constrain women’s participation, whereas the presence of additional female laborers in the household may help free up individual time. Geographical proximity to fishing ports appears to amplify the supportive effects of different livelihood capital types. Married women tend to prioritize livelihood stability. In contrast, single or divorced women, despite potentially lacking physical capital, remain capable of livelihood transformation due to better access to credit. Overall, these findings provide some empirical evidence that may be useful for designing integrated policies aimed at simultaneously addressing gender equity, livelihood diversification, and ecological protection in mangrove-dependent coastal communities.