Introduction

Many studies point out the growing importance of non-agricultural incomes for the development of rural areas, highlighting, in particular, the phenomenon of the pluractivity^a of rural families^b. Arkleton Trust’s research³ on pluriactivity in European agriculture found that the level, composition, and evolution of income of agricultural households depended crucially, between 1987 and 1991, on the regional context (five study areas) and on the situation of pluriactivity (its greater or lesser intensity and diversification). This research showed that the agricultural workforce is increasingly directed to non-agricultural activities and activities related to agriculture outside the establishment, apart from the fact that pluriactivity was shown as a stable phenomenon and disseminated in all types of regions of the European Economic Community: 63 % of agricultural establishments present a member involved with non-agricultural activities.

The World Bank, on their part, produced innumerable reports recognizing that agriculture is not the exclusive activity and, in many countries, not even dominant, in rural areas, emphasizing the importance of non-agricultural income in these areas in developing or underdeveloped countries. According to the World Bank, this would really be an important component of development and of the reduction of rural poverty that should be considered in the actions to be implemented. The main recommendation for rural action strategies is to broaden the focus on the rural definition, abandoning the «strict approach of the agricultural sector»⁴⁾⁽⁵. According to this same understanding, the IBRD⁶ states that «the priority of (rural development) policies is to create more jobs, both in agriculture and in the non-agricultural rural economy».

In Brazil, in the nineties, professors Jose Graziano da Silva and Rodolfo Hoffmann of the Universidade Estadual de Campinas (Unicamp), State of São Paulo) led many investigations that arose in the scope of the Rurban Project. They used the microdata from the Continuous Survey of Homes (in Brazil, National Household Sample Survey, PNAD^c), which contributed to revealing rural economic and social dynamics in Brazil, allowing the dimensioning and characterization of the presence and importance of the non-agricultural activities within the rural residents, farmers and not farmers. This enabled a statistical treatment to a set of behaviors of the different types of families that structure Brazilian rural areas, particularly family agriculture. The visualization in numerical terms enlightened the perception of the importance not only of the family farmers but of the other types of families residing in the rural areas of Brazil, highlighting their economic and social contribution to the development of the country.

The possibility of accompanying and analyzing the behavior, equally important, of the categorization of rural producers over the years, according to the position in the occupation, type of occupation and position in the family, as well as income by family and gender and the sources of income, have helped clarify and reveal the trajectory of Brazilian rurality in a panoramic way. The research that was developed with the methodology of the Rurban Project -especially the use of the typology of families that this project elaborates- helped to reveal substantially the regional differences of the categories of farmers and rural non-farmers, and allowed the deepening of the knowledge about the dynamics and trends of employment, income and living conditions of rural people and families, no longer in a static scenario, but with a move towards a more dynamic analytical framework, that enables an almost continuous analysis over time. The obtained results increased the rigor of the analyses and allowed many interpretations about the reality of Brazilian rurality, and particularly about family agriculture.

The Rurban Project consisted of three phases and began to be developed based on the dissemination made by the IBGE (Brazilian Institute of Geography and Statistics) of the microdata from the Continuous Household Survey, which made it possible to carry out a new characterization of the population living in the Brazilian rural environment and, particularly, of the state of São Paulo. In Phase III of the Rurban Project^d, the previous topics were deepened in the various federal rural and agricultural units of some regions of the country. The studies proved the previous results, revealing that the pluriactive families tend to have higher schooling, a better standard of living, a lower proportion of the elderly and, consequently, a lower proportion of government transfers in the family income. With regard to income, field studies showed the positive impact of pluriactivity on family income, although this is still less than the impact of retirement. Even so, the pluriactivity of self-consumption in the composition of family income was not revealed high as foreseen in the study hypothesis.

In addition, the new activities were analyzed in more detail. Studies on rural industry were undertaken, revealing its growing importance, products, and marketing channels, among other topics. The importance of outsourcing in agricultural activities, in services, especially domestic service and rural tourism, was a subject that was investigated in many regions of the country. An unpublished study also analyzed the environmental impact of some of these new activities. A large contingent of workers residing in the countryside commutes from the rural area every day to work in the urban areas closest to the place where they live. In addition to this phenomenon, it was discovered that a large number of non-agricultural activities are developed in the rural area, such as the installation of small industries for the production of compotes, cheese, and other dairy products, among other activities related to housing, leisure, and provision of services^e. Another movement that the researchers studied was the installation of condominiums in rural areas, which interferes with agricultural activities and urban planning itself. Although secular in the country, these activities recognized in the countryside did not have great economic importance. They were carried out in the yards of the houses, as personal hobbies or small intensive agricultural businesses (fish farming, horticulture, floriculture, «table» fruit growing, raising small animals, etc.), which became important work and income alternatives in the rural environment in the most recent years. This was generically called new rurality^f.

The Rurban Project research produced many reports and publications with important information that contributed to the redefinition of public policies, which began to consider this new existing reality in the country’s rural environment. Despite the existence of such publications with various important subsidies for public policies, there are still many research gaps. One example is that it has not been possible to carry out a complete characterization of agricultural and non-agricultural activities with the microdata from the PNADs / IBGE, mainly in regard to their location since the PNADs do not allow disaggregation levels similar to those of the census investigations.

In other words, the researchers did not have sufficiently detailed and disaggregated information about non-agricultural activities in rural areas with regard to Brazilian municipalities. Such information, when obtained, could be very useful for the verification and understanding, not only of the incidence of these activities but also for the offer of subsidies for future research on more specific topics. Kageyama¹¹ conducted a pioneering analysis that was based on data from the Population Census of the State of São Paulo, which resulted in the reinforcement of the conclusions of previous Rurban Project investigations and showed that, mostly, rural residents in this state had non-agricultural, low-skilled and precarious occupations. That is to say, the result of the study confirmed the conclusions obtained from the analysis of the data from the Continuous Household Survey (PNADs).

The investigations with the microdata from the PNADs have been very important to crystallize not only literature with new concepts but also to guide the public policies for Brazilian rurality. These advances do not free the investigations from constant reflections and revisions from the theoretical and/or empirical point of view. This article proposes, therefore, to proceed with the reflection of this review by analyzing the coherence that the microdata from PNAD has with the reality of rural families, considering that it is not census research, despite the rigor with which the sample is conducted. In other words, the aim of this article, of a methodological nature, is to verify the absolute and relative magnitude of families and pluriactivity in the South Region of Brazil through the comparative analysis of the results of the reprocessing of the microdata from the Census of Population¹ (in Brazil, Demographic Census -DC) of 2010 with those of the Continuous Household Survey² (PNAD) of 2011. Additionally, it is intended to verify the validity/importance of the PNAD for the analysis of the characteristics and evolution of the different types of families living in rural and urban areas and the relevance of the use of these information sources to deepen on rural/urban relations at present.

Materials and Methods

Results and Discussion

Table 1 presents, in a comparative way, the information extracted from the microdata from the 2010 Population Census (2010 DC) and the Continuous Household Survey (PNAD) of 2011 in relation to the type of families analyzed here.

When the relative values in Table 1 are determined, it is possible to notice the proximity between the compared results between the 2010 DC and the 2011 PNAD, practically for all types of families of the adopted typology. This observation is valid both for the two columns that present the relative shares of each family type in the general total of southern families, as well as for the other two columns that show the family distribution within each type of family according to the position in the occupation (employer, self-employed worker or family agriculture, employee).

In this sense, it can be observed in Table 1 that, from the point of view of the absolute numbers, there are differences that can be explained between the results obtained when the 2010 DC and the 2011 PNAD are analyzed. These differences in the results can be explained by 1) different reference periods for information collection of the 2010 DC and the 2011 PNAD, as shown in the methodology section; 2) the fact that the PNAD is a sample investigation and, therefore, produces estimates subject to errors. In this case, it can also be seen in the columns that record the relative/percentage values (Table 1 and following) that these differences are not expressive to the point of invalidating the PNAD because it is a sample; 3) for the simple fact that the year of the DC is 2010 and the year of the PNAD is 2011, the same could be observed between two PNADs, for example, 2011 and 2012.

It is evident in Table 1 that the most important difference to be highlighted in this comparison occurs in the group of employer families, probably due to the small number of existing families, whether in the case of the Population Census data as in the case of the Continuous Household Survey (PNAD)¹⁰. Even in this case, the difference is not expressive enough to consider that PNAD data cannot be used for analysis. In general terms, the information is quite similar and would allow researchers to use it, confirming what was already known for the case of the State of São Paulo in the publication of Kageyama^j, using data from the Population Census.

Table 2 records the information relevant to the different types of families living exclusively in the rural areas of the Southern region. Note that the comments made above on Table 1 are practically valid for the data compared in Table 2, and more explicitly for the pluriactive and non-agricultural families, in the three family groups (employers, self-employed or agricultural families, employees).

The slightly more significant difference between the 2011 PNAD data in relation to the 2010 DC, in the case of agricultural families, is probably due to the fact that the information was collected at different times. It is already known from the article published by Graziano da Silva and Del Grossi¹⁵ that the effect of the different reference periods of the PNADs (year and week) is numerically important, but the authors state that in the case of Brazil the magnitude of this error that was calculated was less than 0.5 % and did not affect significantly the results obtained for the agricultural PEA as a whole.

Excluding the participation of non-agricultural families, both in the general total and in the total of each family group (employers, self-employed or agricultural families, employees), Table 3 offers a comparative analysis between the family agriculture (families of self-employed workers) and non-family agriculture (families of employers and employees). Table 3 shows that both in the 2010 DC and in the 2011 PNAD the families of self-employed workers and those of employees were the vast majority of the universe of southern families linked to any agricultural activity (in the 2010 DC they corresponded to 98.1 % and in 2011 PNAD they reached 94.3 %). That is to say, the families of employers remain with relatively smaller participation in that universe.

But if we observe the distribution of family types according to the area of activity (agricultural family and pluriactive family) within each family group according to the position in the occupation (employing families, families of self-employed workers or family agriculture, employed families) ), we can verify that the family group with the largest participation of pluriactive families is the group of families of employers (65.1 % in the 2010 DC and 52.95 % in the 2011 PNAD), followed by the other two groups that, although with a smaller relative participation of the pluriactive families, still exhibit an expressive participation in this family type, both in the 2010 DC and also in the 2011 PNAD. In addition, as we have observed in Tables 1 and 2, it is also observed in Table 3 that the percentage distributions show small differences between the 2010 DC and the 2011 PNAD data -differences that can be explained by the previously presented reasons-, suggesting the statistical confidence in the use of PNAD for studies of spatial distribution and of the evolution of the different family types from the adopted family typology, particularly, but not exclusively, of pluriactive families.

From the exclusive point of view of the families residing in the southern rural areas, Table 4 reaffirms the observations that have been made about the information recorded in Table 3, but basically with only one visible difference: that the pluriactivity of the employers families showed a significant divergence, that is, in the 2010 DC 52 % of the non-family agriculture of employers was pluriactive, while in the 2011 PNAD this participation fell to 24.1 %. As we have highlighted above, a part of this difference can be explained by the reasons stated at the beginning of this section.

However, it should be noted that this relative reduction is not due to an absolute reduction of the pluriactive families, from the 2010 DC to the 2011 PNAD. On the contrary, since the 2010 DC recorded four thousand southern rural families of pluriactive employers, while the 2011 PNAD recorded six thousand families of this same type. This reduction -from 52.0 % in the 2010 DC to 24.1 % in the 2011 PNAD- is more because of the absolute (and relative) growth of the southern families of agricultural employers.

This same type of observation can be made in Table 3, which collects the information from the total southern family agriculture and non-family agriculture (rural plus urban). Note than in this table, the case of the employing family agriculture, 2011 PNAD recorded expressively more agricultural and pluriactive families than the 2010 DC did. But just as was recorded in Table 4, the most significant difference between the 2011 PNAD and the 2010 DC is in the families of agricultural employers and not in the families of pluriactive employers, which makes the participation of the latter family type lower (in 2011 PNAD) compared to the first family type.

All information presented here show that the analysis of Escher and others¹⁶ may have been somewhat hasty: first because they claim - without any intention to verify as we are proposing in this article, to compare the PNAD with DC - that the PNAD, given that it is a sample investigation (which do not collect information in all municipalities as censuses do), would be limited to studies on pluriactivity; and, secondly, because they highlight the strong presence of pluriactivity in the non-familiar agriculture as information supposedly exclusive of the Agricultural Census of 2006.

Another observation to highlight is that, as we have shown it before, the participation of the non-family group of employers in the universe of southern families is very small, which justifies the emphasis the countless studies produced by the Rurban Project put on the pluriactivity of the family agriculture which, in the 2010 DC and the 2011 PNAD, gathered more than half of southern families, as well as also allowed highlights to pluriactivity of families of employees grouping almost the other half of the southern families (table 3).

Another comment that deserves highlight refers to what is considered non-family agriculture. In the work of Escher and others¹⁶ there is an explicit association with «employers» agriculture and no relation with the families of employees. In the work of these authors, the employees seem not to form families, let alone deserve, as a family group, the attention of the investigators. Obviously this it is not a problem of Escher and others¹⁶, but of the database they used, the Agricultural Census, that did not allow them to consider the group of the families of employees (and the pluriactivity within).In this sense, the greater importance of PNAD is evidenced, more than the 2006 Agricultural Census, for the analysis of pluriactivity, since PNAD includes both agricultural family (self-employed workers) and the two non-familiar agriculture groups (employers and employees). It is a more adequate database for the purpose of studying the pluriactivity as mentioned in the innumerable articles produced by the researchers of the Rurban Project.

On this, both the DC as the PNAD allow us to describe better what could be called non-familiar agriculture. Perhaps the greater difficulty in the use of the data from the Agricultural Census of 2006 for the analysis of the pluriactivity is precisely the subject of the analysis unit, that in the case of the Agricultural Census it is the «Agricultural establishment». In the case of the Population Census, as well as the PNAD, the analysis unit is the «residence», which is a unit closer to the idea of family. This is clear when the Agricultural Census denominates the pluriactivity like «economic pluriactivity», according to what appears in the «technical notes» of the Agricultural Census that the IBGE published¹⁷.

Picture 5 registers the information relative to the absolute and percentage distribution only of the two contingents of agriculture non-familiar (employers and employees), enabling more precise observations on these two groups within what can be considered like non-familiar agriculture.

In Table 5, which includes only two groups of non-family agriculture and its general total, we can observe the overwhelming predominance of non-family agriculture of employees, in the 2010 DC (96.0 %) and in the 2011 PNAD (85.2 %). Although the PNAD works with samples, it is this research, and not the DC, that registers a greater share of the employing non-family agriculture (14.8 versus 4.0 %on the DC). Obviously, this difference can be explained with the reasons that have already been mentioned more of once. But what interests us, in this case, is to call the attention on the fact that the PNAD, unlike the DC, although it does not collect information of all the municipalities it is absolutely not less reliable than the DC. In this sense, the last two columns of Picture 5 reveal that the results extracted from the microdata from the PNAD are compatible not only with the results of the DC but also with the ones Escher and others¹⁶ found in the Agricultural Census^k.

As one might expect, considering only the non-family agriculture that resides in the southern rural areas (Table 6), the concentration of non-family agriculture used is even higher in the 2010 DC it was 97.8 % and in the 2011 PNAD was 88.8 % - compared to the same numbers in Table 5. The analysis of the last two columns of Table 6 was made when comments were made in Table 4.

Conclusions

The presented article has fundamentally a methodological nature, as stated in its objective. But it is possible to make some considerations so that this initial treatment of the data can have continuity through future investigations.

There is no doubt that since the publication of the 2006 Agricultural Census the new methodologies have allowed new research incursions in what refers to the studies on agricultural establishments, mainly because they show the existence of non-agricultural activities in their interior and/or the practice of non-agricultural activities outside the establishment by the heads and/or other members of the establishment.

But according to the specificity of the analysis presented here and contrary to what other authors have argued rather hastily, both the information from the microdata from the PNADs and those of the Population Census are extremely valid for the studies that have been developed within the Rurban Project, as well as in others. These two databases are totally valid for a thorough review of rural-urban relations in Brazil and, through their territorial divisions, to rediscuss the relevance of the biunivocal relationship between rural and agricultural and urban and non-agricultural, which means making the necessary break with rural-urban and agricultural-non-agricultural dichotomies.

The election of the Agricultural Census or the PNAD for the study of the pluriactivity must not have the statistical significance of one of these databases as a criterion since it is understood that it has been possible to show in this analyses that the PNAD is statistically compatible with the Population Census and the Agricultural Census. Therefore, choosing criteria for one or the other database must meet the objective of the investigation. That is, for the purposes of the research carried out by the Rurban Project, the PNAD was and remains the most appropriate; as well as, on the other hand, for studies of another nature, such as investigating the types of production and marketing of the different pluriactive establishments, the Agricultural Census is undoubtedly the most recommended one.

It is repeated, finally, that both in the Population Census and in the PNAD the unit of analysis for the study of the pluriactivity is the «residence», which is a unit closer to the idea of family. However, as regards to the information on the production within the establishments of the pluriactive families (or strictly agricultural), we agree that the Agricultural Census is more suitable for a study of such character.