Drivers of Livelihood Strategies: Evidence from Mexico’s Indigenous Rural Households

Abstract

Mexico has the largest Indigenous population in the Americas and the most native languages in the region. These Indigenous peoples face a similar set of structural barriers to achieving more sustainable livelihoods, including criminal violence and huge hurdles to accumulating assets, in addition to their poverty. The purpose of this paper is to identify the main drivers of sustainable livelihood strategies in Indigenous households in rural Mexico. Using cluster analysis and a multinomial logit model based on a mixed approach that employs a traditional perspective of development alongside the sustainable livelihoods approach (SLA), our results suggest that a significant proportion of Indigenous households engage in livelihoods linked to the environment and nature, while only a small segment of households has been able to accumulate assets and adopt more profitable non-farming livelihoods. Our findings also reveal how the creation of human capital, the provision of basic services, and support to mitigate the effects of extreme weather all contribute to reducing risk for Indigenous households. The findings suggest that public policies must target these specific issues in order to solve structural problems that limit the efficacy with which Indigenous households use their family assets.

1. Introduction

Indigenous peoples account for 5% (370 million) of the global population, with a significant proportion (15%) living in extreme poverty [1]. Mexico has the highest concentration of this population in Latin America [2]. According to the World Bank [3], the number of Indigenous people in Latin America is not easily estimated. One of the most important reasons is the absence of accurate, reliable information, as well as the very nature of Indigenous identities. A good example of this is that only ten countries in the region have included ethnic information in their household surveys. According to the 2020 Mexican census, 6.1% of Mexicans (7,364,645 individuals) self-identify as Indigenous [4]. Common denominators among this population are geographic isolation, poverty levels, and life in small rural localities of less than 2500 inhabitants [4]. In 2018, the percentage of Indigenous people living in poverty in Mexico was much higher (69.5%) than the percentage of the non-Indigenous population (39%) [5]. On one hand, Indigenous livelihoods are a consequence of their limited access to public goods and services, and their inability to develop their abilities and capabilities [6,7]. On the other, their local context exposes them to a variety of risks, such as the loss of family crops and assets due to extreme weather events [8,9]. In addition, there is convincing evidence that for many years this population has endured discrimination and social exclusion [7].

In recent years, the violence of criminal groups has represented another problem for the livelihoods of the Mexican rural sector [10]. The proliferation of criminal organizations has damaged rural Indigenous livelihoods, mainly due to constant disputes over territory [11,12]. These multi-stressors erode the Indigenous population’s quality of life and often drive them to activities that inhibit their social and economic development [2,13].

In a context of rural poverty, Indigenous households seek to generate their livelihood strategies not only to obtain higher monetary income but also to minimize their exposure to risk (illness of family members, changes in local prices, extreme weather events, and civil violence and conflict). To make these life choices, they consider their assets and the contextual factors of local vulnerability. Following the IPCC’s definition of vulnerability, this research understands vulnerability as: “A present inability to cope with external pressures or changes, such as changing climate conditions. Contextual vulnerability is a characteristic of social and ecological systems generated by multiple factors and processes” [14] that affect them [8,15].

Several empirical studies have examined the determinants of sustainable livelihoods in households around the world [16,17,18,19,20,21]. Some research analyzes metadata from various countries [20,21], while another set of projects has carried out case studies [22,23,24]. Recent literature about the topic suggests that the livelihoods of rural households in low- and middle-income countries are closely linked to the environment and nature [8,20]. These studies also argue that an additional element to consider as a determinant of livelihoods is associated with weather disturbances (droughts, rains, and extreme temperature) that modify the survival strategies of people in rural areas [9,25].

Despite the significance and wide diversity of livelihoods among Indigenous peoples in developing countries, most academic literature has focused on studying the link between access to assets and the livelihoods of non-Indigenous groups [13,15,26], leaving the Indigenous dimension poorly studied [7,20]. An exception is the studies by Lin and Polsky [27], Tschakert et al. [19], and Komey [28], who recognize the significant heterogeneity of rural Indigenous households both in terms of vulnerabilities and access to capital, as well as their potential to participate in a variety of livelihoods. Understanding how rural Indigenous households can progress into more sustainable livelihood strategies is a complicated issue since they comprise a largely heterogeneous population in terms of language, culture, ecosystem, access to assets, and livelihoods [7,29].

Aimed at addressing this gap in the literature, the goal of this article is to identify the characteristics of local vulnerability, household, and community that function as primary drivers of livelihood strategies for a sample of Indigenous households in rural Mexico. The questions guiding this research are: (1) What are the livelihood strategies of Indigenous households in Mexico’s rural areas? (2) Which household assets and local vulnerability enable Indigenous households to make the transition from low- to high-return livelihoods? The answer to these questions is key to understanding the barriers that these households face when attempting to develop their abilities and adopt more sustainable livelihood strategies in developing countries with a large Indigenous presence.

Our research makes three fundamental contributions to the debate on the determinants of Indigenous livelihoods. First, the pioneering development studies on Mexico approached the topic by considering that rural households seek to maximize monetary income [30,31,32]; however, in recent years, scholars have identified that household assets and the local vulnerability of Indigenous communities produce multiple stressors that affect their livelihood coping strategies [9,20,33]. Despite considerable literature on the topic, national studies to date have not combined diverse approaches to analyze the assets and local context of rural Indigenous households in a disaggregated way. This study aims to take a first step in this direction.

Second, international organizations and academic literature on development have provided evidence that non-farming activities play an increasingly crucial role in the income-generating activities of rural households and communities in developing countries [30,34,35]. However, this conclusion may be biased in terms of the place occupied by household agriculture in poor Indigenous households [29,36]. In Mexico, seven out of ten Indigenous households work on family farms with an average size of 5 hectares, growing primarily corn [37]. Therefore, the disaggregated microeconomic data used in the current study will help us to identify more accurately the continued relevance of agricultural activities for income generation in Mexico’s rural Indigenous households.

Third, while rural poverty in Latin America has decreased by more than 2.7% in just over a decade, dropping from 22.7% to 20% between 2010 and 2018 [38], this trend has not been observed in Mexico’s Indigenous population. Between 2008 and 2018, the percentage of Indigenous people living in poverty in rural areas went from 79.1% to 78.7%, remaining almost unchanged in percentage terms, but increasing in absolute terms [5]. These figures show that social development policies in Mexico have so far been unsuccessful. This is due in part to the fact that Mexico’s public policy has not considered the structural factors that could foster the development of Indigenous people’s abilities and capacities in allowing them to participate successfully in sustainable development [39]. This study aims to provide insight into the difficulties Indigenous peoples encounter in entering labor markets. This last point is essential for the creation of public policies that can effectively foster rural development.

This research focuses on the family assets that enable Mexico’s rural Indigenous households to transition from low-return to high-return economic activities. Using the sustainable livelihoods approach (SLA) as a framework and data from Mexico’s National Rural Households Survey (ENHRUM), we conducted a cluster analysis and generated a typology of three groups of Indigenous households: (1) Subsistence farming (SF); (2) No off-farm income (NOFI); and (3) Off-farm income (OFI). Based on this classification, we developed an econometric model to analyze the interrelationship between access to family assets and the livelihoods of rural Indigenous households. Overall, the main findings suggest that a small proportion of Indigenous households has accumulated assets to develop profitable non-farming livelihoods. Our results also reveal that human capital, basic services, and support to mitigate the effects of extreme weather all contribute to reducing risk for Indigenous people.

The paper is divided into five sections including the introduction. In Section 2, we present the literature review. The methodological approach is explored in Section 3. The main results and discussion are presented in Section 4. Lastly, the conclusions are drawn in Section 5, along with public policy recommendations.

2. Literature Review

2.1. Determinants of Livelihood Strategies in Rural Households

Scholars such as Ellis [15] and Natarajan et al. [8] have argued that households have different livelihood strategies. For instance, Ellis [15] points out that for rural households, livelihoods represent monetary and nonmonetary activities to ensure survival in the vulnerability context of their local communities. Agricultural production, trade, service provision, and self-employment are some of these productive activities. In addition, the international migration of a household member can also operate as an income-generation strategy to cover the costs of clothing, healthcare, and even education for some household members [8,15,17]. Thus, the amount and quality of capital a household possesses will determine its skills and livelihood strategies. There are five main categories of capital: financial (livestock, earned income, cash savings, jewelry, and pensions); physical (the infrastructure and goods required for production, such as machinery, tools, roads, highways, and irrigation); natural (land, water, flora, and fauna); human (education, training, and health); and social (networks, associations, and relationships of trust). Moreover, these assets are strongly interconnected. While social capital, for example, is key to reducing costs associated with labor markets and can grant access to information and the development of knowledge, human capital can lead to the use of other assets [15].

In 1997, the UK Department for International Development (DFID) suggested that livelihood strategies depend on the local vulnerability context, factors over which households and communities have little or no control in the short term [8]. Some of these factors include the effects of climate change, civil conflict, and human health. To understand the relationship between the five types of capital and the local vulnerability context, we must also explore the influence of institutional structures and processes. Institutional structures can be divided into public organizations (political and judicial bodies) and private organizations (companies, corporations, social organizations, and NGOs), which work together as the mechanism that creates and implements policy and legislation. Institutional structures contribute to the interaction between institutions and households through the institutional processes they establish, which encompass political (regional and sectoral), market, cultural, and power relationships [15].

Studies on rural households have relied on various indicators to determine livelihood strategies, based on two broad approaches. First are those that use expenditures (or income) as the outcome variable to examine the productive activities of rural households [31,32,40]. The second approach argues that risk is the main reason why households decide to adopt different livelihood strategies [23,41,42,43]. According to the DFID [8], a livelihood is sustainable if it can withstand stresses and shocks or if it can recover from them while improving its assets. This second perspective has dominated developmental literature since the late 1990s, underscoring the fact that the quantity and quality of assets allow households to decide what kind of livelihood strategies they will adopt to survive the local vulnerability context [9,15]. Despite these perspectives, the sustainable livelihoods approach (SLA) has recently transformed to face the challenges of the twenty-first century, such as COVID-19 [8].

2.2. Livelihood Strategies of Indigenous Households in Mexico

Mexico has the most native languages in the Americas: 68 languages and 364 registered dialects [2,44]. The rich cultural, agroecological, political, and historical diversity of the territories inhabited by Indigenous people has led to a heterogeneity of the livelihood strategies they depend on for survival [45].

Currently, Mexico’s rural households are experiencing greater penetration by non-farming activities [36]. Indigenous domestic economies have been affected by these dynamics; however, an important difference with regard to non-Indigenous households is the enormous dependency on the land, since family agriculture and natural resources are pillars of their life strategies [29,46]. Aside from these productive activities, government cash transfers play a relevant role in their livelihood strategies [2]. Nonetheless, the strong dependency of their livelihoods on the land means that they are exposed to multiple climate stressors and that their risk management strategies are limited since they own few assets to cope with these phenomena [9,47].

Although the number of studies addressing livelihoods has increased globally, few have analyzed the Indigenous dimension [48]. Research focused on Indigenous livelihoods in Mexico predominantly involves case studies [22,36,49,50,51]. The majority of this literature has investigated how the participation of Mayan communities in public policy influences the creation of policies that boost sustainable development [50,51,52]. Similarly, other studies, such as that by Groenewald and Bulte [49], have explored the positive impact of social capital on livelihood strategies in Popoluca households. The determinants of ecotourism livelihoods have recently been studied. Ávila-Foucat et al. [22] have analyzed the effects of households’ capital assets on livelihoods on coastal communities in Oaxaca, Mexico.

3. Methodological Approach

3.1. Conceptual Framework

As we previously mentioned, rural Indigenous households in Mexico exist in heterogeneous social, cultural, and environmental conditions, with different livelihoods, risks, and local vulnerability [52,53]. Nevertheless, some assets appear to have more relevance than others in a household’s livelihood strategies, regardless of the context. The choice of livelihood strategies is determined by five main factors: (1) family labor and education, (2) land area, (3) migration networks, (4) environmental conditions, and (5) access to markets. In most cases, the role of these variables has been explored through aggregate studies [30,32,54]. However, depending on the household type and local vulnerability context, each of these variables will play a different role. For instance, a person’s identification as Indigenous is directly linked to family labor since some characteristics of Indigenous households can inhibit participation in paid activities. This is partly because labor markets are incomplete or nonexistent in rural communities, effectively preventing community members, and particularly Indigenous peoples, from participating in non-farming activities [30,35].

To analyze livelihood strategies, family labor should be considered together with the sociodemographic characteristics (health, schooling, and training) of working-age household members [40,55]. For example, Tacconi et al. [21] and Härtl et al. [56] argue that rural households with livelihoods highly linked to nature are normally more intensive in family labor in their production processes. According to Bhandari [41], De Janvry and Sadoulet [30], and Choithani et al. [17], there is a strong link between years of schooling and household participation in non-farming livelihood strategies.

Among rural households’ assets, land ownership stands out [35,40,54], given that through land, households can gain access to other benefits, such as monetary resources, livestock, and credit. A recent study examining livelihood diversification in eight countries in sub-Saharan Africa notes that access to credit and land are two fundamental factors that explain why some households can access livelihoods with higher returns [20]. Indeed, land ownership is a type of savings that can essentially protect households from stressors, while also allowing them to manage risks [35]. According to empirical studies, greater access to land means less low-wage employment (such as construction) [30]. Other research argues that when an additional ha of land is acquired, household participation in agriculture and livestock raising grows, while participation in paid labor markets decreases [32].

The significance of social networks in rural livelihoods and poverty has been extensively documented in migration literature [17], and particularly the role of remittances in encouraging credit applications and increasing the demand for local goods and services in rural areas [32,57]. In addition, remittances may help cover basic expenses when a household experiences local vulnerability ranging from accidents and family illness to extreme climate events [47]. Recent studies have documented that in rural contexts, where livelihoods are scarce, social networks offer an opportunity to engage in livelihoods that are more profitable than agricultural production [17,20].

Rural Indigenous households are highly exposed to local vulnerability and economic shocks that can deeply impact their everyday life, including events ranging from long-term disturbances (family accidents or loss of assets) to food insecurity due to climate change [9,55]. Climate-related events can cause considerable damage to food security and dramatically reduce household income when households are dependent on climate [21,58]. Scholars agree that community development (mainly identified through access to well-maintained roads) can result in higher returns on family capital. For example, households with similar assets located in different contexts can follow contrasting paths [59]. The present study explores how infrastructure like roads can better connect rural and semi-urban communities, thereby increasing access to markets and employment opportunities. Moreover, a higher frequency of extreme weather conditions can increase local vulnerability levels and decrease the possibility of earning a higher income.

3.2. Data Source

The data used to carry out this study were taken from the 2007 Mexican National Rural Household Survey (ENHRUM) (For more details on the sample design and conceptual aspects of ENHRUM, see http://bdsocial.inmujeres.gob.mx/, accessed on 10 June 2022). Mexico’s national information and census office (INEGI) carried out the sampling frame. Its design is complex (probabilistic, stratified, multi-stage) and the ultimate unit of study is the household. The survey interviewed a total of 1543 households in 80 rural communities across Mexico, of which 21% are Indigenous. Communities in the sample have between 500 and 2500 inhabitants and are located in fourteen states in five rural regions (see Figure 1). ENHRUM was originally designed as a panel survey to capture changes in Mexico’s rural economy, yet public safety issues have prevented re-interviewing rural households in the sample regions. Likewise, one of the potential limitations of ENHRUM is that it was not applied in all states having a high Indigenous presence. For the purposes of this paper, we considered only Indigenous people aged six years and older living in these communities. In ENHRUM, ‘Indigenous’ refers to a person who speaks an Indigenous language—that is, individuals who answered “yes” to the question “Does (NAME) speak an Indigenous language?” This survey does not contain information related to Indigenous self-determination due to their cultural heritage, an aspect that could expand the identification for Indigenous people in our study. The survey included two components: the Community Survey (CS) and the Household Survey (HhS).

The CS was used concurrently with the HhS in the 80 communities surveyed. The questionnaire was aimed at individuals who were properly informed on the dynamics of their community and the local vulnerability context. Key respondents included local representatives, school principals, and heads of health-service organizations. Variables were primarily measured at the local level, allowing their use in estimating the local vulnerability context. These variables include the most common climate-related events, changes observed in the livelihood strategies of both local residents and migrants, village-level characteristics, and infrastructure. The rural communities analyzed are characterized by heterogeneous livelihood strategies, which range from livestock raising and crop production (for household consumption and for sale in markets), to the use of local natural resources. Other options within and outside the community are commercial services and paid non-farming activities.

The HhS was the main survey used to collect sociodemographic data and information on Indigenous households’ assets. Households were selected as the unit of analysis and defined as a group of one or more persons who usually reside in the same dwelling, spend jointly on food, and who may or may not be related [60]. Accordingly, the HhS collected information on households’ economic and social characteristics, including sociodemographic data, land holdings, access to credit, income from livelihood strategies (both farming and non-farming), migration, use of natural resources, and exposure to damage caused by natural phenomena. Information on nonmonetary activities was also collected, such as the use of family labor and production for household consumption, a key element given the type of livelihood strategies used by Indigenous peoples in Mexico.

3.3. Empirical Strategy

Our empirical strategy is based on an innovative three-stage analysis. The first stage classified households based on their livelihood strategies by looking at income-generating activities and the local vulnerability context. A livelihood was defined as a combination of strategies that allow households to survive and manage risks [35]. The criteria used for classification were: sector criteria (agriculture, industry, and services; see Barrett et al. [31], spatial criteria (local, national, and international), and functional criteria (self-employment and paid employment). Lastly, the classification was based on seven activities: (1) agricultural paid labor; (2) non-farming paid labor (construction, industry, education services, and transport); (3) family self-employment (grocery shops, restaurants, auto-repair shops, stationery stores, and sewing shops); (4) crop production; (5) livestock; (6) use of natural resources; and (7) migration (internal and international).

Once the classification process was complete, we calculated the households’ annual net income [34]. Subsequently, households were divided into deciles based on net income. Lastly, we conducted a cluster analysis with several variables: physical capital (value of agricultural and non-farming physical assets); human capital (number of household members, type of leadership); income (deciles); financial capital (value of formal and informal credit); natural capital (land size); social networks (number of people who could help in case of emergency); and local vulnerability context (average distance to large regional centers and index of damages due to droughts and hurricanes). It should be noted that the above list covers a variety of measuring scales, including ordinal, cardinal, and dichotomous variables. To address this issue, we used a dissimilarity measure with a group number equal to three (K = 3) and the Gower formula to create a distance (dissimilarity) between these variables [34].

For the second stage, a statistical analysis of the livelihood-related variables used in stage one was conducted, allowing us to examine the differences in livelihood strategies among study groups as well as their local vulnerability context. In the third and final stage, we applied a multinomial logit model to determine Indigenous households’ main drivers of livelihood strategies. In this econometric model, the dependent variables were the livelihood strategies derived from the first stage.

To assess whether a household was correctly classified based on its strategies and income-generating activities, a Hausman test was applied to ensure that the independence of irrelevant alternatives (IIA) hypothesis holds in our econometric model. The test indicated that the estimators in the multinomial logit model are nonbiased and consistent; therefore, the household groups’ specifications were deemed appropriate. These results are shown as coefficients in terms of the relative risk ratio (RRR). Generally, these results are interpreted by assuming a unit change in one of the independent variables, while average values are maintained for all independent variables. Following this logic, an increase in the probability of taking part in livelihood strategies is represented by values greater than 1, while a decrease in this probability is represented by values between 0 and 1.

In the multinomial logit model, the explanatory variables were the assets and local vulnerability context. Accordingly, education and household labor were used to account for human capital. A household’s ownership of farmland was selected as a proxy for natural capital. Because ENHRUM collects data on migratory networks, social capital was measured using social networks. Lastly, the local vulnerability context was determined through market access, infrastructure, and damages from climate-related events.

Table 1. Definitions and metrics used to describe the characteristics of Indigenous livelihood strategies.

Variable	Metrics
Net annual income	Annual net income (MXN 2007)
Physical features of the house and housing services
Rooms	Total number of rooms in a house
Wood or sheet metal walls	Walls in the house (1 = wood or sheet metal walls; 0 = otherwise)
Concrete roof	Composition of the roof of the house (1 = concrete roofs; 0 = otherwise)
Latrines for bathrooms	Type of bathrooms in the house (1 = latrines; 0 = otherwise)
Landline	House has access to a landline (1 = landline; 0 = otherwise)
Use of wood for cooking	House uses wood for cooking (1 = uses wood for cooking; 0 = otherwise)
Homes without a refrigerator	House without a refrigerator (1 = with a refrigerator; 0 = otherwise)
Characteristics of the household head
Age	Age of household head (in years)
Years of schooling	Years of schooling of the household head
Characteristics of the household
Hh size	Total number of people living in the household
Hh labor size (members aged 11 to 65 years)	The average number of members aged 11 to 65 years
Years of schooling of workers in the Hh	The average number of years of schooling completed by household members aged 11 to 65 years
Household owns farmland	Household has access to owned farmland (1 = yes; 0 = otherwise)
Household income activities
Households with income from non-farming wages	Paid non-farming activities (1 = yes; 0 = otherwise)
Households with income from agricultural wages	Paid agricultural activities (1 = yes; 0 = otherwise)
Households that depend on self-employment	Self-employment activities (1 = yes; 0 = otherwise)
Households with livestock holdings	Livestock holdings (1 = yes; 0 = otherwise)
Households that use natural resources	Use and collection of natural resources (1 = yes; 0 = otherwise)
Households that participate in crop production	Crop production (1 = yes; 0 = otherwise)
Sources of external income: public and private
Households with PROCAMPO	Household receives government cash transfers from PROCAMPO (1 = yes; 0 = otherwise)
Households with OPORTUNIDADES a	Household receives government cash transfers from OPORTUNIDADES (1 = yes; 0 = otherwise)
Households with domestic remittances	The household receives domestic remittances (1 = yes; 0 = otherwise)
Households with remittances from the United States	The household receives US remittances (1 = yes; 0 = otherwise)
Market access and transaction
Distance to large urban centers	Average distance from the village to main population centers (km)
Available credit	Annual formal or informal credit available (MXN 2007)
Vulnerability to extreme events
Index of damage due to climate-related events	Level of exposure to extreme events that cause loss of family agricultural assets (droughts, freezes, hailstorms, and strong winds) (percentage)

Source: the authors, based on ENHRUM survey 2007. ^a: The goal of this program was to reduce poverty in Mexico (details in Yörük et al. [2] and Novotny et al. [25]).

provides details regarding all the variables included in this analysis (definitions and metrics used by the aggregation level and category).

4. Results and Discussion

4.1. Descriptive Statistics: Contrasts in Indigenous Rural Households

Considering the SLA, the results of the cluster analysis generate a typology of three groups of households. These outcomes represent dependent variables in the multinomial logit model: (1) Subsistence farming (SF), (2) No off-farm income (NOFI), and (3) Off-farm income (OFI). The first group, Subsistence Farming (SF), includes Indigenous households that produce on a small scale with a farming system of subsistence agriculture. Household crops are rainfed (corn, beans, and squash) and cultivated on the smallest plots of land in the entire sample, with an average area of 4.6 ha. Livelihood strategies in this sector are also based on raising animals (chickens, pigs, goats, and sheep) and using natural resources (75% of these households make use of firewood, wild fruits, rabbits, iguanas, and birds, among other natural resources). Farmland and home gardens are important because they produce a variety of foods to survive. A recent study in Mexico found that 18 animal species and 212 plant species were used by Indigenous households [61]. These households also rely heavily on government cash transfers (PROCAMPO and OPORTUNIDADES). In general, they have limited monetary resources, with an average annual net income of MXN 9821 (less than USD 500). Such features make this group very sensitive to changes in the prices of agricultural inputs and food, as well as to extreme natural events. Their access to credit is very limited: MXN 631 per year. On average, households in this group consist of eight members living in two-room houses. In addition, in 61% of these homes, the walls are made of wood or sheet metal, and only 21% have poured concrete roofs. The homes have limited services indoors: 58% out of the households use latrines as bathrooms, 80% burn firewood for cooking, and 69% do not have refrigerators for food storage. Another notable feature is that the household decision-maker has an average age of 54 and has usually completed an average of 3.5 years of schooling. The labor force (ages 11 to 65) has the fewest years of schooling in the entire sample, at 4.7 years (

Table 2. Main sociodemographic characteristics of Indigenous households’ livelihood strategies.

	Livelihood Strategies of Indigenous Households
Variable	Total (n = 286)		Group 1: SF (n = 91)		Group 2: NOFI (n = 102)		Group 3: OFI (n = 93)
	mean	sd	mean	sd	mean	sd	mean	sd
Net annual income (MXN)	34,045.3	3316.4	9821.1	8971.3	23,857.0	10,235.5	68,922.9	35,467.5
Physical features of the house and housing services
Number of rooms	2.49	1.41	2.40	1.30	2.37	1.45	2.72	1.44
Wood or sheet metal walls (%)	0.46	0.49	0.61	0.48	0.47	0.50	0.31	0.46
Concrete roof (%)	0.29	0.45	0.21	0.41	0.22	0.41	0.44	0.49
Latrines for bathrooms (%)	0.53	0.49	0.58	0.49	0.56	0.49	0.45	0.50
Landline (%)	0.19	0.39	0.14	0.35	0.21	0.41	0.22	0.42
Use of wood for cooking (%)	0.71	0.45	0.80	0.40	0.72	0.44	0.61	0.48
Homes with refrigerator (%)	0.44	0.49	0.31	0.46	0.43	0.49	0.56	0.49
Characteristics of the household head
Age	52.28	13.79	53.73	15.66	50.98	13.56	52.29	11.95
Years of schooling	3.76	3.35	3.54	2.83	3.54	2.92	4.22	4.16
Characteristics of the household (Hh)
Household owns farmland	0.65	0.47	0.83	0.37	0.61	0.48	0.52	0.50
Hh labor (members aged 11 to 65 years)	3.19	1.99	2.68	2.04	3.00	1.79	3.90	1.95
Years of schooling of workers in the Hh (%)	5.85	2.91	4.73	2.88	5.76	2.87	7.03	2.56
Land size (has)	4.97	8.58	4.64	7.27	4.80	7.67	5.49	10.54
Household income activities
Non-farming wages (%)	0.20	0.40	0.06	0.24	0.14	0.35	0.41	0.49
Agricultural wages (%)	0.37	0.48	0.10	0.31	0.57	0.49	0.40	0.49
Self-employment (%)	0.22	0.41	0.18	0.39	0.16	0.37	0.32	0.46
Livestock holdings (%)	0.71	0.45	0.82	0.38	0.64	0.48	0.67	0.46
Natural resources (%)	0.71	0.45	0.75	0.43	0.76	0.42	0.62	0.48
Crop production (%)	0.70	0.45	0.84	0.36	0.69	0.46	0.58	0.49
Sources of external income: public and private
PROCAMPO (%)	0.27	0.44	0.39	0.49	0.24	0.43	0.18	0.38
OPORTUNIDADES (%)	0.45	0.49	0.39	0.49	0.47	0.50	0.48	0.50
Domestic remittances (%)	0.25	0.43	0.21	0.41	0.23	0.42	0.30	0.46
Remittances from the United States (%)	0.08	0.28	0.02	0.14	0.13	0.34	0.09	0.29
Market access and transaction costs
Distance to large urban centers (km)	35.49	30.81	36.65	30.99	28.66	26.74	41.84	33.49
Available credit (MXN)	1157.34	4396.76	631.86	2525.24	969.60	2749.21	1877.41	6671.86
Vulnerability to extreme events
Index of damage due to climate-related events	58.57	33.91	62.27	31.36	55.76	34.86	58.037	35.26

Source: Authors’ estimates using data from ENHRUM survey 2007.

).

The second group, No Off-farm Income (NOFI), in addition to its reliance on land-based activities (agriculture and use of natural resources), is the segment that is most active in the agricultural labor market, as its members are major suppliers of local paid labor. Since most agricultural production in rural areas is seasonal, an important part of the household income remains stable for several months but can fluctuate considerably for the rest of the year. The average size of owned farmland in this group is 4.8 hectares. The annual net income of these households is MXN 23,857 (less than USD 1200), far greater than that of Group 1 (SF). However, in general the characteristics and services of these households are very similar to those of the previous group (see Table 2).

In the third group, Off-farm Income (OFI), self-employment is significant in households. Their livelihood strategies include selling goods (in grocery stores, stationery shops, tortilla shops, or as catalog merchants selling cosmetics and other products), and services (auto-repair shops and sewing shops, among others). Statistics indicate that these are the most prosperous Indigenous households in the rural sector; their dwellings have three rooms and unlike other households, four out of ten have poured concrete roofs. One in five households have a landline and 61% burn wood for cooking. They have the most highly educated labor force (seven years of schooling). Additionally, 41% of these households have members who participate in the non-farming labor market (see Table 2). Many of them depend on domestic remittances as well. They have also accumulated more farmland (5.4 ha) than the previous groups.

4.2. Econometric Findings: Determinants of Livelihood Strategies of Indigenous Rural Households

Table 3. Multinomial logit model used to predict the effects of Indigenous family assets and the vulnerability context on Indigenous households’ livelihood strategies, reference category Group 1: SF (n = 91).

Variable	Group 2: NOFI (102)		Group 3: OFI (n = 93)
	RRR a	SE b	RRR	SE
	[1]	[2]	[3]	[4]
Characteristics of the household head
Young household head (ages 20 to 39 years)	1.294955	1.224632	1.27201	1.294919
Adult household head (ages 40 to 59 years)	0.502201	0.215717	0.733889	0.354871
Elderly household head (60+ years)	0.604705	0.252713	0.996930	0.456288
Household head with elementary education or less	0.817247	0.505156	0.483710	0.303184
Household characteristics
Hh labor (ages 11 to 65 years)	0.988807	0.100512	1.29548 ***	0.137978
Years of schooling of family workforce	1.112587	0.075959	1.312143 ***	0.102391
Migratory networks
Domestic remittances	1.000115 **	0.000058	1.000163 ***	0.000058
International remittances	1.000131	0.000087	1.000153 *	0.000088
Natural capital
Household owns farmland	1.241310 ***	0.009222	1.155966 ***	0.0064879
Variables concerning vulnerability
Community is in a municipal seat	4.895737 ***	2.66565	2.608106	1.716892
Index of damage due to climate-related events	0.987904 **	0.005790	0.993192	0.006151
Homes with a landline	2.095503 *	0.936347	1.872148	0.913273
Constant	3.504708	3.058766	0.456084	0.446725
Log likelihood	−255.2348
Chi-square	117.22
Prob > chi-square	0.0000
Pseudo R2	0.1868

^a: RRR (Relative Risk Ratio); ^b: SE (Standard error); Single, double, and triple asterisks (*, **, ***) indicate statistical significance at the 10%, 5%, and 1% level. Source: the authors, based on ENHRUM survey 2007.

shows the results of the multinomial logit model where Group 1 (SF) households are the category of reference. The table (columns 1 and 3) presents the estimated RRR and standard error (SE) (columns 2 and 4). The results illustrate the impact of household assets (human capital, access to owned farmland, migration networks) and the local vulnerability context on the probability of belonging to the categories of Indigenous households described in the previous section: SF, NOFI, and OFI.

Our findings indicate that the variable of young, adult, and elderly heads of household is not statistically significant in Groups 2 (NOFI) and 3 (OFI). The RRR decreases for young to elderly household heads. For example, the corresponding relative risk ratio coefficients for OFI are 1.27201 and 0.996930, respectively. Regarding household heads with an elementary school education or less, it is interesting to note that this is not a statistically significant variable, although the signs and magnitude of coefficients in each of the household categories are as expected (0.817247 and 0.483710). These results suggest that the characteristics of household heads are not relevant in explaining the transition from low-income to high-income livelihoods. This is due to the fact that within Indigenous culture, individuals commonly acquire social commitments, such as marriage and parenthood, at an early age. Thus, household heads are not able to continue their studies and quickly enter local labor markets, which hinders their access to higher-income livelihoods [16]. Similarly, in rural Mexico it is common for older household heads to have lower schooling levels and to carry out farming activities, while younger family members usually build new homes and seek new opportunities in non-farming markets [32,62].

The RRR coefficient of availability of working-age labor (ages 11 to 65) increases with higher levels of household income, although this indicator is significant only for Group 3 (OFI), with a relative risk ratio coefficient of 1.3. It is also interesting to note that the RRR changes for each of the categories examined (NOFI and OFI) as the family workforce increases its educational level. The findings indicate that the educational level of the labor force is key to explaining the economic returns from livelihood strategies—a result consistent with previous research [15,35]. In this sense, we could argue that the probability of belonging to the reference class diminishes at a greater rate as the available household workforce becomes more highly educated. This means that the probability of participating in non-farming labor markets increases and consequently, the family’s opportunities to earn greater revenues increase [32,47]. Nevertheless, our results contrast with authors such as Lay et al. [62], who pointed out that the additional impact of education is limited when related to seizing opportunities in marginalized locations. In contrast, recent research indicates that there is not sufficiently clear evidence of the correlation between the availability of working-age labor and livelihoods, because while more working-age labor helps to diversify livelihoods, it is also true that the households require greater crop production for their food [21,56].

Regarding migration networks, we found that domestic remittances are significant for Groups 2 (NOFI) and 3 (OFI), with RRR slightly higher than 1 in both cases. Likewise, international remittances are significant for Group 3 (OFI). The above illustrates the fact that remittances from national labor markets increase the probability of leaving the base category (SF). Meanwhile, international labor markets offer opportunities for only a small segment of Indigenous households—those with more family assets [47,57]. These results point in the same direction as a recent study for India that shows the rapid process of national (rural–urban) and international migration, which offers different opportunities to rural households depending on their access to assets [17].

The variable of households that own farmland was significant for Groups 2 (NOFI) and 3 (OFI). The resulting coefficient indicates that land ownership reduces the probability of belonging to the reference category of households (SF). This result highlights the importance of land access as a way for Mexico’s rural Indigenous households to diversify and create the conditions needed to reduce their vulnerability. Although the intensity may vary throughout the nation, a central characteristic of Indigenous peoples is their deep and crucial connection to plots of land and natural resources. Indigenous communities have integral systems where the production of dryland crops, livestock, and the collection of resources are essential for their food security [11,29]. Therefore, among Indigenous groups, land ownership ensures part of their food security and increases the likelihood of accessing other off-farm productive activities [62,63]. Proof of the importance of farmland to Indigenous households is the existence of 1.1 million Indigenous production units in rural Mexico. These properties account for 10.6% (20.8 million) of the nation’s total hectares, yet only 32.9% is cultivable [37]. Indigenous households, therefore, must participate in other livelihoods.

Lastly, we observed that access to domestic telephone lines is relevant for the OFI group. Access to this service increases the probability of leaving the base category, possibly because landlines at home may lead to a decrease in transaction costs, taking into account that many Indigenous peoples are located in remote locations [42]. According to information from the EC, on average, Indigenous communities are 3.5 km farther from urban centers than non-Indigenous communities. Therefore, domestic telephone service can enable Indigenous households to obtain valuable information that allows for their insertion into rural–urban migration networks and that facilitates their participation in daily and productive activities [47,64].

As for vulnerability context variables, belonging to a community within a municipal seat is a significant variable for the NOFI group only. This finding suggests that many municipal seats in rural Indigenous areas lack the required infrastructure and offer reduced opportunities to work in non-farming activities (such as municipal services, for example); the local supply of non-farming employment offers few job opportunities, which do not necessarily require groups with family assets to have special abilities. Similarly, closeness to municipal seats facilitates the sale of farm goods, making farming activities more profitable for the NOFI group [40,65]. Some authors argue that proximity to cities or participation in developed regional markets increases crop yields due to the widespread use of agricultural inputs as compared to remote locations [42]. This may explain why living in a municipal seat fosters household participation in the production of cash crops and in self-employment activities. We observed similar findings for the index of damage due to climate events, a variable of significance for the NOFI group. According to EC data, the above finding means that surveyed rural localities in our sample have precarious community development and that weather events increase the probability of belonging to the SF group since they narrow the possibilities of ensuring food security. However, Jezeer et al. [9] found no empirical evidence to support the idea that climate risks are a determinant of the livelihoods of Indigenous communities.

5. Conclusions

This research provides important insights into the relationship among sustainable livelihoods, vulnerability to local context, and disposable assets in rural Indigenous households. Given that Mexico has the largest Indigenous population in the Americas and the literature has focused on studying the livelihoods of non-Indigenous groups, this study’s findings contribute to the understanding of how Indigenous households might transition into higher-income livelihood strategies while preserving those traits specific to their culture and their traditional ways of integrating, along with the communal aspects that characterize them.

By employing a mixed approach combining essential elements of the SLA with a traditional development perspective, we were able to capture the heterogeneity prevailing in Mexico’s rural Indigenous households and build a typology of three Indigenous households: (1) Group 1 (SF); (2) Group 2 (NOFI); and (3) Group 3 (OFI). Based on these groups, we developed a multinomial logistic econometric model that can help identify the interrelationships among access to assets, the vulnerability context, and the livelihood strategies typical of these households.

Our results reveal that the livelihood strategies of a significant proportion of households (SF and NOFI) consist of an integral system of activities linked to the environment and nature. Similarly, a small segment of households (OFI) has been able to accumulate assets and participate in more profitable non-farming livelihoods. On the other hand, specific results derived from the econometric model reveal how access to the creation of human capital (expressed through more years of schooling), the provision of basic services such as landlines, and support to mitigate climate events all contribute to the transition from low- to high-performance livelihoods; that is, not belonging to households in the SF group. These factors also accelerate the process for participating in local non-farming labor markets and having better opportunities to improve agricultural yields.

These findings disclose the need to solve structural problems that limit the efficacy with which Indigenous households use their assets (including access to basic public services such as medical care, social security networks, and access to telecommunication networks, along with housing, credit, and adequate roads to regional development centers). Therefore, Mexico requires a long-term integrated policy to develop Indigenous regions by strengthening rural households’ economic system and the sustainable use of their natural resources, from an innovative perspective that will allow them to face the challenges of the twenty-first century.

Even though the results obtained in this research are statistically reliable, it should be noted that our analysis presents limitations that could provide the basis for future studies. First, due to the enormous heterogeneity of Indigenous peoples in Mexico, it would be highly desirable to conduct more disaggregated regional analyses that cover municipalities with a high Indigenous presence. The absence of details and up-to-date information for each of these areas represents a significant barrier, a pending task that could be completed through the development of specific projects. Second, our econometric model does not explicitly capture whether households diversify because they wish to minimize the risk of local vulnerability or because they want to maximize their monetary income. Some of our study findings suggest that because of the limited capabilities of SF compared with OFI households, the SF group may want to minimize risk, while the OFI group takes advantage of the limited opportunities offered by non-farming jobs in the rural sector. To address the above point, deeper analysis is needed along with additional data sources.

The enormous complexity of Indigenous groups in terms of lifestyle and the threats to their traditions and culture poses significant challenges. New research initiatives need to incorporate multi-disciplinary approaches and use complementary collaborative frameworks in the quest to understand the difficulties Indigenous peoples face in developing countries.