Social inequality

Inequality refers to a non-equitable distribution of opportunities or wealth generated in a society. Social inequality is the asymmetrical distribution of goods and services such as health, education and income between households, social classes, gender, and races in a society (^{Conapo, 2005}).

The concept of “Inequality is broader than poverty, insofar as it is defined based on the total of a given distribution, and not only on a part of the income distribution of individuals (or households), which is defined to be under a certain poverty line” (^{Rodríguez, 2008}). This is independent of the distribution mean and, in its place, the second moment is addressed exclusively: distribution dispersion (^{Litchfield, 1999}).

Social inequality is manifested in problems such as the deterioration of income, the depreciation of work, and the low coverage of basic social services. The inequality in income distribution, specifically, has an effect on employment, education, health, housing, and public and residential space, and it segments social life.

The study of Welfare requires addressing multiple historical, cultural and political factors, in addition to the economic. Among the latter, income distribution between all components of the social group (family or household) is taken into account. However, these three concepts belong to the same family and in general the indicators used to measure them are compound (^{Haughton and Khandker, 2009}).

In the literature about social inequality the indicators of inequality can be positive measures, which do not refer explicitly to any concept of social welfare and the regulations, based on a function of welfare. To the first group belong the statistical indexes traditionally used to analyze the dispersion of a distribution of frequencies such as variance, standard deviation, coefficient of variation, and analysis of quantiles. Meanwhile, to the second group belong those that are based on a function of social welfare such as the Gini index, the Lorenz curve, and the Theil index (^{Haughton and Khandker, 2009}). In this study we estimate the inequality with the Gini and Theil indexes.

Concept of income and consumption

In México, since the mandate of the Sustainable Rural Development Law (Ley de Desarrollo Rural Sustentable, LDRS), to measure the social welfare of the population, a pattern of consumption that reflects a minimum of welfare is used and whose market price constitutes the poverty line (^{CONEVAL, 2015}). Thus, the households whose income is below this are considered to be income poor. The incidence of income poverty is an indicator that is broadly used worldwide which reflects the percentage of the population in a certain scope that does not have the sufficient monetary resources to acquire the goods and services considered as necessary in their social environment.

With the purpose of making the results from this research comparable, the concept of income that ^{INEGI (2015)} proposes was used, and which is the same that CONEVAL uses in the measurements of welfare and poverty in México. The concept of income is formed by the monetary and non-monetary entries that satisfy three criteria: regularity, availability and patrimony. According to INEGI, the current income is made up of the following five categories: income from work; property rental; transferences; estimation of housing rent; and other current incomes (^{INEGI, 2010}). Given the characteristics of this study and because of cost issues, it is not possible to attribute a value for housing rent, or property rental, which is why for simplicity they will not be taken into account for the income estimations.

The welfare situation of rural households in México and Puebla

The state of Puebla is the fifth most populated state in México, with 5.78 million inhabitants and the ninth in contribution to the Gross Domestic Product (^{INEGI, 2010}). However, it is also one of the entities with greatest social backwardness and income inequality; it occupies the fourth place, after Guerrero, Oaxaca and Chiapas. By 2014, the population in situation of poverty and extreme poverty was 64.5 % and 16.2 %, respectively. The population without access to welfare was 69.7 % (^{CONEVAL, 2015}).

Regarding the monetary income, the most frequently way of knowing its distribution is using an inequality measure, such as the Gini or Theil coefficients. For the state of Puebla, the Gini index went from 0.48 in 2010 to 0.57 in 2014, situation that should be addressed for this index decreased at the national level; it went from 0.50 to 0.48 in the same period; another aspect that accounts for the situation of social backwardness in the state is food insecurity, which was 26.2 % in 2014, and has remained almost without change in recent years, moving from 26.4 % in 2010 to 25.2% in 2015 (^{CONEVAL, 2014}; ²⁰¹⁵).

The situation of poverty and inequality could be explained by a series of shortages that limit the opportunities of the population, such as lags in education, social security, and health services, which were 25.3 %, 72.3 % and 40.4 %, respectively, in 2010. By 2015, the two first shortages did not improve in a relevant manner, with values of 22.9 % and 71.3 %, respectively, while the lack of access to health services was 18.9 % in this last year.

In Puebla, at the municipal level, the situation of income and its distribution presents important contrasts, for while in 2014 Chichiquila presented 96.3 % of its inhabitants in situation of asset poverty, Santa María Tlaltempan had 3 % in this condition, which in relative terms is good and places it as one of the municipalities with lowest incidence at the national level.

In terms of the study zone, the municipality of Tlapanalá has a population of 7063 inhabitants, with a Human Development Index (HDI) of 0.805; the Gini index is 0.38 and 80.9 % of the population is in asset poverty. San Salvador el Verde, with a population of 24 812, has HDI of 0.815; Gini index, of 0.426; 69.5 % of the population in asset poverty; and 13.5 % in extreme poverty. Finally, Coatzingo had 2,714 inhabitants; HDI, of 0.7502; Gini index, of 0.4313; 76.6 % of the population in asset poverty, and 30.8 % in extreme poverty (Table 1).

The municipalities under study also present different productive and natural conditions. San Salvador el Verde obtains most of its agricultural and livestock income from the production of grains and porcine livestock, Tlapanalá from vegetables and bovine livestock, and Coatzingo from vegetables and bovine livestock (Table 2).

According to this, the countryside has a relevant role in the reproduction of households, because the agricultural and livestock income is an important source of food security and monetary income, much more so for San Salvador el Verde and Tlapanalá and, to a lesser degree, for Coatzingo and Tlapanala.

The hypothesis addressed in this study is that the social welfare of rural households, calculated based on the income and the consumption, is explained by sociodemographic, productive characteristics and by the environmental conditions under which they live. Thus, the objective is to calculate and analyze the behavior of indicators of socioeconomic welfare, their distribution, and their relation with the sociodemographic and productive characteristics of rural households.

Location of the study area

The municipality of San Salvador el Verde is located in the central-west part of the state of Puebla. Its geographic coordinates are parallels 19° 12’18’’ and 19° 21’ 54’’ N and meridians 98° 26’ 54’’ and 98° 93’ 18’’ W. The transition from temperate climates in Valle de Puebla to semi-cold in the low parts of Sierra Nevada is present in the municipality. The mean annual temperature is 12-18 °C. The temperature of the coldest month is between -3 and 18 degrees. The highest zones of the Iztaccíhuatl are covered in snow and there are high-mountain meadows around them; the lower slopes of the Sierra Nevada are devoted mostly to rainfed agriculture. To the east, there are zones of irrigation agriculture that are part of the large irrigation zone that surrounds San Martín Texmelucan and the largest from Valle de Puebla.

The municipality of Tlapanalá is located in the southwestern part of the state of Puebla. Its geographical coordinates are parallels 18° 38’ 24’’ and 18° 44’ 06’’ N and meridians 98° 28’ 18’’ and 98° 35’ 36’’ W. The municipality has a flat topography in general with a soft descent in the northeast-southwest direction, which ranges from 1500 to 1300 m of altitude. The municipality is located within the zone of warm climates in Valle de Izúcar de Matamoros, with two identified: semi-warm and sub-humid warm. The greater part of the municipality presents areas devoted to agricultural activities; there are large rainfed agricultural zones. The mountainous areas are covered with low deciduous forest associated to secondary tree and shrub vegetation.

The municipality of Coatzingo is located in the central-southern part of the state of Puebla. Its coordinates are parallels 18° 31’ 36’’ and 18° 39’ 36’’ N and meridians 98° 08’ 18’’ and 98° 14’ 60’’ W. It presents a single climate of warm type: sub-humid with summer rains. A large part of its territory is covered with low deciduous forest; its location coincides with that of Litosol and Regosol soils, and both with the zones that present a certain slope in their relief. There is an important area devoted to irrigation agriculture in the flat zone; intermediate between the irrigation zones; and in the areas of low deciduous forest there are rainfed agriculture zones that cover a considerable area.

A sample was built by conglomerates in two stages. This method was selected because there is no framework for household sampling, although the communities within each one of the municipalities selected were known. The primary units were formed by municipalities, and the secondary units by households, inside each municipality. This is a probabilistic sampling design, whose principal characteristic is that the population is divided into natural groups; in this case, municipalities and communities that we call conglomerates.

In the first stage, three municipalities with probabilities proportional to their size were selected; in the second, a simple random sampling was carried out. For the random selection of households, each population units was divided into quadrants and in each one the predetermined number of households was distributed. The pilot survey and its final application were carried out in September and November 2014, with the participation of four professionals with socioeconomic training who were previously trained. In this stage, three actions were carried out simultaneously; the surveyors were trained, the pilot survey was applied, the structured questionnaire was corrected, and the structured questionnaire was applied, face to face, to heads of family. To understand in greater detail aspects of consumption habits and expenditure throughout the year, two in-depth interviews were performed in each municipality, with key informants; reliable heads of household, with whom the researchers have kept working relations for several years.

The municipalities selected for the study are San Salvador el Verde, Tlapanalá and Coatzingo, and the main selection criterion was the contrast detected in field visits, in their productive and sociodemographic characteristics. Table 3 presents the household sample and their distribution.

Questionnaire design and data analysis method

The questionnaire was made up of 10 sections: sociodemographic and housing characteristics; agricultural and livestock productive activities; commerce and services; transferences, migration, financing; natural and unforeseen resources; household consumption; organization, commercialization and other services; and food security. The measuring scales were nominal, ordinal and rational, with a total of 157 questions.

Measuring the income and the inequality

Social welfare was calculated based on the concepts of absolute and relative poverty (^{Foster, 1998}). For the first, the poverty lines approach was used, defined as income lower than the threshold income required to satisfy basic needs. The other welfare measure is the measurement of relative poverty, or inequality, which is defined as the way in which wealth is distributed in a society, that is, how much a person receives in comparison to other members of society. Inequality can be measured for any distribution in a population, but when speaking of economic inequality, it refers to the distribution of consumption or income in the population (^{Foster, 1998}). In this study we estimate inequality in income and in consumption.

There are many methods for measuring inequality in income and consumption, each one with its characteristics and limitations. The most frequently used measure is the Gini index. The coefficient varies between zero and one. A Gini coefficient of zero reflects full equality, while a Gini coefficient of 1 shows perfect inequality (^{Medina, 2001}). Graphically, the Gini index can be represented as the area between the Lorenz curve of the income distribution and a hypothetical equality line. The Lorenz curve traces the participation in the income accumulated versus the population distribution. In the case of perfect equality, the Lorenz curve would be the same as the equality line, which makes the Gini index zero. The Gini coefficient is capable of providing a basis for the value judgement of income inequality. The internationally accepted standard is that if the Gini coefficient is lower than 0.2, the income distribution observed is highly similar; if the value varies between 0.2 and 0.3, the income distribution is relatively average; if it is between 0.3 and 0.4, the income distribution is moderately high; and if it is higher than 0.4, the income inequality tends to be large (^{You Hongbing, 1998}).

The second measure of inequality is the entropic Theil index (^{Theil, 1967}), which derives from the notion of entropy of the information theory; it is a finer concentration measurement than the Gini coefficient. The Theil index is normalized in an interval between zero and one, where values close to zero mean lower inequality, and close to one higher inequality. This indicator gives greater weighting to the transferences performed to individuals or households with low levels of income, indicating that there is a greater reduction of inequality in face of a transference towards a low stratum than towards a middle one, and fulfills the axiom of additive decomposition; that is, a transference from a richer person to a poorer one makes the index decrease.

The Gini coefficient and the Theil index fulfill the properties of good indicators that measure income concentration, which justifies their use in this study. Mainly, they satisfy two conditions that are desirable in the estimation of inequality; scale independence, that is, that there shouldn’t be variation in face of proportional changes due to the values of the variable or changes in scale. The Pigou-Dalton condition or weak transference principle means that the inequality must be reduced if there is transference of income from a rich household to a poor one. There is a third condition that only the Theil index fulfills, and it is called strong transference principle or condition of relative change. This criterion demands that the concentration measure has a greater decrease in inequality if the income redistribution is made from a rich household to a poor one, than if the transference is made from a rich household to one of middle income (^{Medina, 2001}).

For the data analysis, in addition to the calculation of the Gini and Theil indexes, following the methodology proposed by ^{Haughton and Khandker (2009)}, analysis of variance was used to compare the mean of some variables related to poverty and inequality, between municipalities, and the multiple regression analysis was used to explain the behavior of income of households.

Inequality

The distribution of the income and consumption in the municipalities studied presents important differences. The Gini and Theil indexes estimated a higher inequality for San Salvador el Verde and a lower one for Tlapanalá and Coatzingo. It is important to note that in this order, San Salvador el Verde, Tlapanalá and Coatzingo present a higher to lower index in poverty and food insecurity (^{CONEVAL, 2015}).

In the case of México it is important to mention that, in general, the states and municipalities with higher levels of income poverty are also those with highest levels of inequality. The high levels of inequality in our country have important implications in its development, for they limit the impact of economic growth in the reduction of poverty (^{De Ferranti et al., 2004}).

This situation is accompanied by important restrictions for poverty reduction, since high levels of inequality prevent a better redistribution of the resources generated by the economy and, therefore, it is necessary to implement specific public policies that lead to the modification of the distribution of resources in these states and municipalities.

Explanatory factors of the income

The household income is one of the most frequently used ways of understanding the ability to obtain goods and services, such as foods, health, recreation and education, among others. The results from the income equation, in this study, reveal the importance of variables both of human capital and productive type. The contribution to the explanation of household incomes from the sample, in order of importance, is: productive assets, schooling of the man or woman head of household, number of livestock heads, and the head of household’s occupation (Table 8).

In the same direction of the results mentioned before, the productive assets of the households (land, transportation means, machinery and equipment) have a positive and significant effect on the household income (^{Yúnez and Meléndez, 2007}). Concerning education, ^{Reardon et al. (2001)} reported that this is a variable that is related with the selection of non-agricultural and non-livestock activities, and also with higher levels of income in the rural households.

An intervention that tends to improve the situation of poverty and inequality in the municipalities studied should consider increasing the educational level and the endowment of productive assets of the households, as well as the role of the non-agricultural and non-livestock activities. These non-agricultural and non-livestock activities play an important role in improving the income of rural households. Regarding this point, ^{De Janvry and Sadoulet (2001)} find that the participation of the non-agricultural and non-livestock income in the total of rural households in México is more than 50 % and that it is associated to higher educational levels.

To have confidence in the results from the regression model, in addition to observing satisfactory values of the F and R² statistics, we tested the model’s assumption; normal distribution of the model’s errors, autocorrelation between residuals (DW), and collinearity (an explanatory variable is not a linear combination of other independent variables). For the normality of the errors, the Shapiro-Wilk test was performed. The p-value calculated was 0.52, which indicates that we cannot reject that the errors are distributed normally. The autocorrelation was tested with the Durbin-Watson statistics, which had a value of 1.969, which is why we reject autocorrelation between the independent variables. Finally, the average Variance Inflation Factor (VIF) statistics for the group of explanatory variables was 1.199, which is why we reject that some of the explanatory variables are a linear combination.