INTRODUCTION

At the moment, people living in rural areas represent 48% of the global population and 20% in Latin America. In this region, poverty has remained stable over the past three decades, proof of which is that 63% of inhabitants live on less than two dollars a day (^{Cepal et al., 2003}). Rural households draw on an array of strategies to survive, such as productive diversification, migration, and intensifying production, all while facing a broad range of risks (Ellis, 2000).

In particular, diversification is a fundamental strategy for rural households (^{Berhaus et al, 2007}; ^{Alobo, 2015}). In Latin America, between 20% and 30% of jobs are non-agricultural and account for 40% of rural income (^{Reardon et al., 2001}), although this figure may vary depending on the context or household assets (^{Mushonagh and Scoones, 2012}). It bears mention that in Latin America, the debate on the role of the productive strategies used in rural households is nothing new. The first studies date back to the nineteen-seventies, when researchers provided extensive analysis of rural economies, emphasizing the importance of land surface area and fertility, as well as family composition (^{Chayanov, 1974}, ^{Stavenhagen, 1977}; ^{Shanin, 1978}; ^{Appendini et al., 1983}).

Later, at the end of the nineteen-nineties, the literature began to shift towards a sustainable livelihoods approach (SLA), which has been used to describe household strategies and the contexts of vulnerability they face (^{Ansoms and McKay, 2010}). The merits of the SLA approach include its recognition of the diversity of livelihoods, the relevance of the social and institutional context, and the role that assets play (^{Alobo, 2015}).

That is why, recently, the approach has been applied to many different countries, including Madagascar, Bulgaria, Guatemala, Peru, Philippines, Nigeria, Mozambique, China, Nepal, and Mexico (^{Winters et al., 2002}; ^{Winters et al., 2009}; ^{Morse et al., 2009}; ^{Giesbert and Schindler, 2012}; ^{Orencio and Fujii, 2013}; ^{Bhandari, 2013}; ^{Hinojosa, 2013}; ^{Fang et al., 2014}; ^{Ávila-Foucat, Saad and Fierros, 2014}). These studies draw on a wide range of assets, although they differ across climatic, cultural, historical, and institutional conditions, and describe different stressors (^{Winters et al., 2009}; ^{Ávila-Foucat et al., 2014}). The effects of stressors on households can vary depending on the characteristics of each household, and can alter daily life temporarily or cause profound commotion that requires governmental intervention (^{Van Den Berg, 2010}; ^{McDowell and Hess, 2012}).

In Mexico, approximately 25 million people (representing 22.2% of the total population) live in rural areas, and six of every ten people (61.6%) who live in rural areas are poor (^{Coneval, 2012}). As a survival strategy, households have diversified, and non-agricultural and non-livestock activities now represent 49.2% of total income (^{Cerón and Yúnez-Naude, 2015}).

The SLA approach in Mexico has only been used in case studies, but not at the national level (^{Ávila-Foucat, 2012}). Nevertheless, these studies have analyzed assets, revealing vast heterogeneity in terms of access to assets and in income (^{De Janvry and Sadoulet, 2001} ; ^{Reardo et al., 2001} ; ^{Yúnez-Naude, 2010}).

Accordingly, the objective of this paper is to identify, using cluster analyses, the different types of rural households in Mexico, considering their assets, monetary income levels, and other aspects related to the vulnerability they face.

THEORETICAL FRAMEWORK

METHODOLOGY

Method to Create the Typology of Households

To construct the household groups, we used multivariate cluster analysis with four variables: 1) predominant source of income; 2) income level; 3) assets; and 4) context of vulnerability. Because the variables used in the cluster analysis are measured with different scales and units! a similar mixed method, with four groups (k=4), and the Grower estimate to create distance between the variables were used (^{Lerner et al., 2013}). Analysis of variance (ANOVA) statistics were also used to ensure that the household groups created were mutually exclusive (Soitani et al., 2012). The procedures used to estimate income sources, income level, assets, and context of vulnerability are described below.

Income sources and activities

First, the income sources were divided following the criteria in ^{Barret and Reardon (2000)}: by spatial terms (local, regional, and national), functional terms (self-employment and wage workers), and sector. Based on these categories, the following seven sources were found: 1) agriculture, 2) livestock, 3) natural resources, 4) agriculture and livestock wage work, 5) non-agriculture and non-livestock jobs, 6) self-employment and 7) remittances.

Second, net income for each household was estimated and production functions for each productive activity were created. Because natural resources and farm products are consumed by members of the households themselves (^{López-Felman, 2014}), and on occasion exchanged with friends or relatives, people were asked how much money they would have earned if the products had been sold locally. Net total income is an estimate of all productive activities and remittances.

Assets and Context of Vulnerability

The assets were classified pursuant to Ellis (2000): human, physical, financial, natural, and social capital. Table 1 shows the variables used to describe the assets and the context of vulnerability.

Table 1 Variables of tbe assets used in household profiles 

Source: Created by the authors based on data from ENHRUM 2007.

RESULTS

Typology of Households

The household groups obtained from the cluster analysis were as follows: smallholder farmers, farm wage-earning households, non-farm wage-earning households, and family business owners. All of the household categories are diversified, so their names are meant only to reflect the main source of income (see Table 2).

Table 2 Makeup of Income and Net Annual Income (Pesos) 

Source: Created by the authors based on data from ENHRUM 2007.

Smallholder farmers

The first cluster includes 20% of the sample (286 households) and is in the lowest end of the income distribution (deciles I and II), with approximately 3,390 pesos a year. Half of their income comes from farm work, natural resources, and remittances from abroad.

Farm wage-workers

The second group includes 30% of the sample (deciles lll-V), and their net annual income is approximately 20,000 pesos. Although most of their income comes from farm activities, one significant difference with respect to the above group is that this group is fairly dependent on the local labor market, because 38.5% of their income is reliant on farm wages and 23% on non-farm jobs.

Non-farm wage-earners

The third group contains 40% of the sample (deciles VI-IX) and their main source of income is non-farming wages (41 %), fundamentally from temporary jobs in construction, transportation, and retail. The households in this group travel on a daily basis to the communities where they work in nearby towns, earning on average 60,000 pesos a year.

Family business owners

The fourth group includes 10% of the sample and their net annual income was a little over 191,000 pesos a year. This set is in the upper income distribution (decile X) and is reliant in large part on non-farming wages (28.6%), self-employment (19%), and also agriculture (20%). These households own the largest swaths of land and have enough capital to invest in family businesses.

In short, it is clear that the first group depends on natural capital (land and natural resources) and remittances, the second two (2 and 3) survive thanks to wages (farming and non-farming), and a fourth lives off of self-employment and is less reliant on agriculture.

Livelihood Strategies

The analysis shows that practically all households do use diversification as a subsistence strategy, as 50% of household income comes from various economic activities. Productive intensification is not an essential strategy in and of itself, as households do not invest in agriculture and livestock infrastructure, and prefer to divide their time among other activities. Remittances are clearly a means of livelihood for the first group, but not for the rest, because income from remittances hovers between 15 and 19%. Accordingly, it is interesting to note that households draw on a suite of strategies at the same time, with a certain degree of emphasis on some strategies that depend on the income and assets they have, as described below.

Human assets: The results show that 25% of smallholder farmers do not have family members in the working age range (11 to 65 years old). By contrast, 77.5% of family business owners have three to six family members in the working age range.

Figure 1 shows level of schooling of the workforce (people aged 11 to 65 years old) for each household group. The smallholder farmer workforce is low-skilled, as 64% have completed only elementary school. By contrast, in the other three clusters, 50% of the family members have some secondary schooling. In addition, 3.5% of the smallholder farmers and 26% of family business owners have completed high school.

Source: Created by the authors.

Figure 1 Education of the Labor Force (11 to 65 Years Old), by Household Group 

A high percentage of smallholder farmers are indigenous (31.8%), while only 7% of indigenous households belong to the group of family business owners. Poverty is tied to being indigenous, isolation, and low levels of schooling.

Physical assets: The composition of physical assets is very similar across the four household groups, given that non-agriculture and non-livestock assets are not the highest-value assets, even for farm wage-earning households. Livestock herds are another major source of investment. In other words, rural households do not invest in agriculture and livestock production; in their place, they prefer diversification or the purchase of livestock, suggesting that agricultural intensification is not the prime strategy (see Table 3).

Table 3 Value of Physical Assets (Pesos) 

Source: Created by the authors based on data from ENHRUM 2007.

Natural capital: All household groups collect firewood that they use as fuel to prepare food. Of all of the natural resources, firewood represents at least 52% of the resources used across all four household groups. Agave, the prickly pear, and other cactuses are the second-most important natural resource group, with the exception of family business owners, who do not use them. Medicinal plants are also important, and at least 5.6% of households use them. Wild mushrooms are important in the diet of smallholder farmers and, surprisingly, wildlife is essential to small business owners (see Table 4).

Table 4 Use of Natural Resources (Percentage) 

Source: Created by the authors based on data from ENHRUM 2007.

On average, the vast majority of household groups has less than two hectares of land, which means that they cannot even produce enough food to survive all year (^{Cepal, 1982}). Smallholder farmers have the largest percentage of land, between two and five hectares, and family business owners tend to have larger plots (five hectares or bigger).

Financial assets: In general, amounts are low and rise in relation to income. The amount of formal credit is highest among smallholder farmers and non-farm wage-earning households, and the majority of informal loans go to farm wage-earning households and family business owners Social networks: Findings show that at least 17% of households have members that have immigrated to the United States, and 11.8% have members that have moved to other parts of Mexico. It also emerged that at least one person would be able to make a loan to other households in case of emergency. It is also clear that smallholder farmers can get more money in case of contingency situations, for an amount of 2,500 pesos, while family business owners would be able to get only half that amount (1,193 pesos).

Context of vulnerability:Table 5 shows that households with a greater share of non-agriculture income are close to regional development hubs, giving them greater job opportunities, while smallholder farmers, who are more reliant on natural resources, are more isolated. Thus, diversification is tied to proximity to cities and access to regional labor markets. Likewise, all rural households have suffered climate-related damage, due to hurricanes and droughts. Similarly, there are no major differences in terms of access to roadway infrastructure.

Table 5 Variables of Vulnerability and Effects on Households. 

Source: Created by the authors based on data from ENHRUM 2007.

In short, assets exhibit similar conditions in terms of access to primary education, investment, natural resources and land size, informal loans, and migration. Damage from hurricanes and drought are similar across all households, but there are differences pursuant to proximity to urban hubs, the highest levels of schooling, and access to loans and credit.

DISCUSSION

The spatial, functional, and sectoral criteria used by ^{Barret and Reardon (2000)} were useful in identifying the rural household profiles. The results show that remittances are an international source of income (spatial criteria), wages and self-employment represent a functional difference, and the agricultural and non-agricultural sectors reflect the main sectoral difference. Net income also has an influence on the profiles and elucidates the inequalities of rural Mexican households.

Diversification is the main livelihood strategy for low-income households, where farm wage-earning households focus their efforts on agriculture and livestock activities, while other household groups are concentrated on non-agriculture and non-livestock activities (except the smallholder farmers). Diversification in smallholder farmers is meant for remittances. In the other household groups, diversification tends towards non-agriculture wages or self-employment. The importance of wages is a sign of the reliance on activities derived from use of land and commerce. This means that diversification requires a minimum amount of income and asset accumulation, as indicated in previous studies (^{Ávila-Foucat, 2012}), and also depends on access to labor markets. In effect, family business owners live the closest (29 kilometers away) to cities, giving them easy access to markets and reducing transaction costs.

These results match other studies that have demonstrated the importance of infrastructure in livelihood decisions (^{Finan et al., 2005}; ^{Jansen et al., 2006}; ^{Ansoms and McKay, 2010}). The household groups obtained in this study are consistent with the findings of other research on the rural economy in Mexico, where agriculture and non-agriculture wages account for 49.2% of total rural household income (^{Cerón and Yúnez-Naud, 2015}). It has also been shown that migration and remittances play an important role in poor communities, but are not enough to spur development (Stefani, 2011; ^{Cohen and Rodriguez, 2005}). In fact, it is hard to say, based on these results alone, that non-agriculture and non-livestock income has fueled rural economic development or has mitigated extreme poverty (^{World Bank, 2008}). ^{Scoones (1998)} defines productive intensification as a better use of natural resources, like land, wildlife, or forests.

Because cross-section data was used in this paper, it is difficult to analyze the process of economic intensification in detail. However, the results suggest that it is not an essential strategy, as households do not invest in agriculture and livestock infrastructure, signaling that agriculture is not a priority in the sample of households used. This finding is because irrigation-driven production is concentrated in specific districts and states throughout Mexico, such as Sinaloa. In summary, livelihood strategies are similar across households, but there is a different emphasis on agriculture and non-agriculture activities and in the categories where assets play an important role in these variations.

The precarious conditions ailing rural households can in part be attributed to limited access to credit (^{De Janvry and Sadoulet, 2011} ). Smallholder farmers and non-farm wage-earning households apply for formal loans in amounts less than 4,000 pesos, likely from local institutions. By contrast, informal loans are used by family business owners, who look for higher loan amounts. All of this spotlights the challenges involved in accessing large formal loans.

Nevertheless, the social networks analysis reveals that all households would be able to go to someone else for help in case of an emergency; what is surprising is that smallholder farmers can access higher loan amounts than higher-income households. Namely, the poorest households are more reliant on their social networks (close friends and relatives) than on formal credit. Other papers have demonstrated cooperation between households (^{Barbieri and Mahoney, 2009}; ^{Mushongah and Scoones, 2012}); however, the findings of this study suggest that cooperation happens through loans and not only by interchanging assets.

Education is a widely-analyzed variable in Mexico, and its importance for development and diversification has been demonstrated in empirical studies (^{Yúnez-Naude and Taylor, 2001} ; ^{Bandhari et al., 2013}). The results show that all of the households have low levels of schooling, but that secondary education is more common in non-farm wage-earning households. Inequalities in schooling are also possibly due to distance from urban hubs and unequal access to labor markets.

Family business owners have the greatest endowment of physical capital, which means that they have the highest potential to invest, which gives them a deeper tool chest with which to confront stress factors. The livestock herd and non-agriculture and non-livestock infrastructure are the most important investments that all of the households make. The livestock herd is an important source of savings and families resort to it in case of problems (^{Berhaus et al., 2007}). Investments in cars and other non-agriculture and non-livestock investments are mainly in transportation needs, because public transportation in rural zones is sparse (^{Adato et al., 2006}; ^{Giesbert and Schindler, 2012}).

Agriculture and livestock infrastructure has a significant percentage share of the investments made by family business-owning households, in spite of farming not being their main economic activity. The minimum plot size for subsistence is four hectares, according to the ^{ECLAC (1982)}, and as is the case for the majority of households in the sample, it is unsurprising that diversification into non-farming activities is the main strategy, implying that this diversification is a way to get out of poverty.

Natural resources are an important income source (Shana and Sundriyal, 2012) and a significant component of the diet (^{Belcher et al., 2005}). other studies have demonstrated that natural resources are important to survival, especially for the poorest households (^{López-Felman, 2014}), but it is interesting to note that the benefits that nature gives to rural households is independent from whether they participate in agriculture and livestock activities or not. Dependence on firewood is due to poverty, but also to culture and culinary traditions. In Mexico, 27.8% of power for households, stores, and public roads comes from wood kindling (^{Sener, 2011}). Medicinal plants are used to cure disease and wild mushrooms are part of the diet for the first three household groups. The wealthiest households hunt wildlife (like bush meet), likely due to the cost of the firearms necessary for hunting.

Family business owners live closest to the cities (29 kilometers away) as compared to smallholder farmers, who face higher transaction costs and challenges in accessing regional markets.

Looking at climate change-related hurricanes and droughts, it is clear that all of the household categories are affected (^{Van Der Berg, 2010}), but vulnerability is exacerbated in certain cases when families have limited resources. For example, smallholder farmers are the most vulnerable because they face the greatest challenges in accessing regional markets and have the lowest education levels. These households also have few agriculture and livestock physical assets (1,411 pesos) to augment productivity or their supply of agriculture and livestock products (milk, eggs, and crops). Moreover, the vast majority of these households have lands (less than two hectares) that are seasonally vulnerable to climate change and suffer from limited access to credit (1,293 pesos).

In Mexico, recent case studies have analyzed livelihoods and vulnerability (Eakin, 2005; ^{Robles-Zavala, 2010}; ^{Giesbert and Schindler, 2012}; ^{Winters et al., 2009}). However, few studies have examined the relationship between income and family assets (^{Winters et al., 2002}). Moreover, most of the research on livelihood strategies has drawn on case studies, and few have provided nationwide empirical evidence.

The results of this study can be used to understand the regional heterogeneity of income, assets, and the contexts of vulnerabilities facing communities, providing a holistic overview to maximize the livelihood strategies that rural households adopt and the decisions they make (^{Carter and Barrett, 2006}; ^{Giesbert and Schindler, 2012}). The relevance of education, age of the workforce, and size of land has also been addressed in the literature, but natural resources, access to informal credit, social networks, and context of vulnerability have received less attention in Mexico with the SLA.

In terms of the limitations of this research, the main limitation is that the survey used was not conducted with an SLA, but it does offer a wealth of data, which make it possible to create proxy variables for each capital, other approaches have been used to analyze productive activities, diversification, poverty, and the rural environment (^{World Bank, 2001} ; ^{Carter and Barret, 2006}; ^{Winters et al., 2009}; ^{FAO, 2013}), but asset-based analysis is the most important tool to understand households' capabilities and skills (^{Winters et al., 2002}). structural aspects, like education and poverty, are undoubtedly crucial to development. However, development policies derived from this study can also be directed at diversification, markets, remittances, credit, and social networks.

Diversification is a major strategy that households use, and sectoral policies have tended to support productive activities, but they are not consistent with each other. Accordingly, in order to support diversification, it would be necessary to have an efficient and coordinated policy that considers sustainable development. Likewise, policies designed for diversification seem to be a better strategy than remittances for rural development, as stefoni (2011) demonstrated.

Access to markets, and especially to jobs, is an important aspect to which close attention should be paid; likewise, household organization may decrease transaction costs (^{FAO, 2013}). For example, household loans must become more accessible in terms of administrative requirements, the guarantees required, and interest rates. Finally, social networks are essential to accessing credit and remittances.

CONCLUSIONS

Based on income availability, family assets, and the context of vulnerability, we identified four household profiles: one tied to natural capital and remittances; another two to agriculture and non-agriculture wages, respectively, and the fourth to self-employment. Productive diversification is the main livelihood choice made in response to exogenous factors and the best development strategy as compared to ramping up remittances. Access to elementary school education, investments, natural resources, size of land, and access to formal loans are all very similar across the different household groups. Likewise, damage from droughts and hurricanes affects every level, but proximity to cities is different, as is level of schooling, and access to loans and credit, structural aspects, like education and poverty, are undoubtedly crucial to development; however, other development policies derived from this study are also important to consider: sustainable diversification, access to markets and commercialization, and access to formal credit.