1 Introduction

The general objective of this chapter is to contribute to comparative analysis of social inequalities based on an analysis of intergenerational social mobility among the member countries of the INCASI network. The intention of this primary comparison is to lay the foundations for further comparisons that will be broader in terms of countries, and more in-depth in terms of the aspects that are currently associated to intergenerational social class mobility.

This goal will be treated in the conventional manner of such studies, addressing absolute social mobility and relative social mobility. A study of absolute mobility is essential in order for the reader to appreciate how it follows on from the previous chapter and to access the descriptive dimension that highlights the different changes in social structure between parents and children in the studied countries. Meanwhile, the study of relative mobility will focus on analysing the association between social class of origin and the social class reached at the time of the survey, thus permitting an examination of the opportunities for accessing different social classes in each country. These opportunities will also be analysed in a comparative manner, to reveal the contrasting possibilities for accessing different social statuses in each country. This involves the use of techniques to monitor the absolute changes observed and for different sample sizes, and to provide parameters for the effects of interest, both in and between all countries.

Our hypothesis is based on a classic idea drawn from the comparative literature on social mobility in late industrialisation countries. According to Ishida and Miwa (2011: 9), there is a relationship between social inequality and the distribution of opportunities: societies that were late to join the industrialisation process generated greater social inequalities and a more uneven distribution of opportunities for social promotion, which led to less social openness. Hence, according to the authors, “the later and rapid the industrial development, the higher the social inequality and the lower the social fluidity”.

This chapter begins with a brief summary of the theoretical aspects of social class mobility, thus presenting the general framework of analysis and situating us within the extensive extant literature on intergenerational mobility. The theoretical perspectives and principal previous literature from European and Latin American countries are reviewed. At this stage, our goal is to generate dialogue between the earliest and contemporary studies from each continent, which serve as a basis and stimulus for our own study, and on which the working hypothesis for our analysis is grounded. In the following section, we present the methodology, discuss the techniques and detail the databases used for each of the countries. In the fourth section, we present the results in two clearly distinguished parts, which focus on absolute and relative social mobility. The former entails a descriptive overview and the latter looks in greater depth at how open or closed societies are, measuring the relative opportunities for people from different social classes. The fifth section summarises the main findings and presents the most relevant conclusions of the analysis. The sixth section contains a discussion of the contributions made by this report and recommendations for the network’s future research programme, as well as pointing out the limitations of the analysis.

2 Theoretical Perspectives and Previous Studies in Europe and Latin America

2.1 Social Mobility in Industrialised Countries

Since the end of World War II, reflection and research on economic growth and development has been linked to social mobility. Improved standards of living became a shared value in the international community, although ways to achieve it differed, given the rivalry between two systems for organising society, capitalism and communism. However, both models shared a focus on the importance of social mobility and public investment in education, the former as result of positive macrosocial changes on an economic and cultural level, indicative of improved life and personal development opportunities, and the latter as driver of access to these opportunities. Hence, achievement was focused on occupational performance and on the redistribution of the means to improve performance in such endeavours.

Twenty-eight years ago, in a study of social mobility, Erikson and Goldthorpe (1992) presented their topography of social theory, which gained major analytical and pedagogical recognition. These authors described three theoretical perspectives on mobility, but for various reasons we can agree to agree that only two of those are still valid today. Together, these two macro-perspectives cover the theory and hypotheses of social mobility: the theoretical tradition of modernisation and the invariance hypothesis called FJH (Featherman, Jones and Hauser).

Sociologists who defended modernisation theory argued that there was a cumulative secular tendency towards development and equal opportunities. Studies like those by Parsons (1951), Kuznets (1955), Kerr et al. (1960), and Treiman (1970) predicted that in the medium and long term, with the social market economy acting as an allocator of resources and opportunities, and with adequate, moderated training and motivation of the workforce, open and unrestricted intergenerational social mobility, equal pay based on meritocracy and remuneration of production factors would be achieved. Different contributions were thereby consolidated, to eventually come together as what was dubbed modernisation theory.

On the other hand, Featherman, Jones and Hauser’s perspective (1975) argued that societies with market economies and nuclear families operated with similar mobility systems, and that inequalities therefore persist across countries. Initially, the national studies by Goldthorpe (1973, 2003), Goldthorpe et al. (1987) led the way in social class mobility studies, and developed the debate on social class mobility in industrial and post-industrial society. And this naturally touched on the kind of development in each country. The counterpoint was the stability or instability of social class intergenerational mobility as a consequence of the different forms of industrial development.

Erikson and Goldthorpe (1992) proposed the theory of constant social fluidity (CnSF) for the case of change over time, or the common social fluidity model, which maintains that relative mobility rates are similar when comparing between countries. Constant fluidity theory is a model for associating between origins and destinations with two variants. On the one hand, it analyses the effect of the cohorts or generations of interviewees on social mobility where the “model of uniform differences” (Unidiff) is identified. And, on the other hand, the Core Model of mobility and social inheritance was identified, emphasising some relationships between social classes related to inheritance, class boundaries, sectoral changes, and short-distance movements between social classes.

Breen (2004) presents evidence of changes in relative opportunities by observing a general decrease in the strength of inequalities when analysing eleven European countries through 117 surveys conducted between 1970 and 1990 that practically cover 60 years of the twentieth century. In terms of absolute mobility, Breen notes two major transitions, the first from agriculture to an industrial society evidenced by the decline in rural classes. Secondly, the shift from industrial society to post-industrial society is observed, as reflected by the increase in classes I + II and III and the decrease in V + VI and VIIa. In terms of relative mobility, Breen is able to observe changes in fluidity in a large number of countries, although this trend is not univocal, since certain countries continue to show constancy, such as men in Great Britain, Ireland and Norway and British and German women.

The findings on the trends in social fluidity are far-reaching, for Breen produces further support for them in a more recent study (Breen et al. 2020; Breen in Salido and Fachelli 2020).

He expands upon classic mobility studies by including the new concept of counterfactual models, an innovation that he developed in 2010 that values the important role of education in intergenerational social mobility trends. The implementation of these techniques leads him to affirm that the expansion of education programmes and more equitable distribution thereof with respect to social origins were the most important factors behind the increase in social fluidity. Breen claims that for the cohorts born in the first half of the century, it is modernisation theory that comes out best, while reproduction theory would best fit the behaviour of cohorts born in the latter part of the century. He therefore maintains that the analysed results cause him to doubt the accuracy of the two rival theories on mobility and proposes the generation a long-term perspective on intergenerational mobility (Breen 2020).

Ishida and Miwa strove to expand the comparative context between countries with different levels of industrial development, and this was one of motivations for our own work. These authors developed a typology that distinguishes early industrialised countries from late industrialised countries. The study coordinated by Ishida in 2008 studies six late industrialisation countries, namely Japan, Taiwan, South Korea, China, Brazil and Chile. A later study (whose findings were reported in 2011 in an unpublished document by Ishida and Miwa) broadens the comparative scope by adding to the six aforesaid countries, those studied by Breen, to also include Israel, Mexico, Italy, Hungary, Ireland, Poland and the United States, eventually totalling 19 countries and concentrating on men aged 30 and 64. The conclusions to their studies showed that there was no historical convergence in terms of social fluidity, that there is a common pattern of fluidity among nations, and that there is convergence in terms of absolute mobility rates. These studies provide the recent background to support the comparative analysis presented herein.

Finally, a very recent study by Hertel and Groh-Samberg (2019) performs an analysis of 39 countries by exploring the relationship between economic inequality and class mobility, whereby the authors are able to confirm their hypothesis that countries with high levels of economic inequality have lower levels of social fluidity.

2.2 Social Mobility and Development in Latin America

Exploration and discussion of social mobility in Latin America has always been associated to studies on development and growth. It almost forms part of the foundations of sociology. But as Solís and Boado (2016) have pointed out, this school of thought was interrupted from the 1970s to 1990s as a result of the political conflicts that affected so many of these societies in that period.

Germani conducted several studies on stratification and social mobility in Argentina. Here, we will refer to the one published in 1963 as an appendix to the Spanish translation of the study by Lipset and Bendix that was published in Buenos Aires. In this article, Germani viewed Argentina as a society that had witnessed a significant amount of social mobility over an extremely long period. There was a first period of major upward social mobility among international immigrantsFootnote 1 as Argentina received about seven million immigrantsFootnote 2 and a second, more recent, period of social mobility among domestic migrants from the rural context to the cities. Germani examined several concepts of social mobility, such as gross (structural and circulatory), demographic and transitional mobility. Gross social mobility is the total amount of social mobility, which can be broken down into two parts, namely the change observed between respondents’ classes of origin and destination (called structural mobility) and circulatory mobility (the difference between gross mobility and structural mobility). Demographic mobility is considered to be dependent on the differential birth rate of social classes, which can generate an over-supply of aspirants to upward mobility. And finally, there is transitional social mobility, which was identified in the substantial changes in the structure of employment, social classes and birth rates that happened in Argentina as it shifted from a predominantly rural, agro-exporting society to an industrial, urban one.

Germani had faith in the transformative power of education, which should provide the support for general social mobility, for it was not only the means with which to provide knowledge to the population, but also to foster the new mentalities that are required in order to attain modernity. That is what had happened in Europe. However, as a consequence of transitional mobility, he warned that in Argentina there may or may not be a correlation between social and educational mobility. This is a matter that concerned him, because a conflict could arise if new class positions were to impose themselves over those promoted by general and meritocratic social mobility, typical of industrialisation. This reflection would accompany him in future years, and from the evidence he collected, he shows that educational mobility did not affect social mobility, so there was no direct correspondence.

Labbens and Solari (1966) examined social class mobility in Montevideo, obtaining several estimates. These authors had fewer secondary sources than Germani to corroborate their evidence, but concluded that Montevidean society had also experienced significant total social mobility. In that year, 44% of Uruguay’s population was concentrated in Montevideo, reflecting the major weight of international immigration and migration from the countryside to the city, as in Buenos Aires. There were large increases in salaried classes at all occupational levels. However, as a result of the stagnation of economic growth based on the export of raw materials from the late 1950s, the authors hypothesised that upward mobility would face obstacles, and in general, social mobility would be reduced in every way.

Solari et al. (1967) understood that education was important for intergenerational social mobility, but that not everyone would achieve social mobility with it, as observed in Montevideo with data from 1959. They noted that an association between education and occupation can produce unexpected results. The relationship between growth and education is not entirely clear. Depending on the geographic contexts of socialisationFootnote 3 they note that education did not always lead to occupations of a similar level. For that period, they highlighted how economic stagnation could lead to an over-educated population, and this will lead to status incongruity, because many people would be unable to find employment in keeping with their level of education.

More recently, Solís and Boado (2016) ran a comparative study on class mobility and stratification in Latin America that brought together researchers from Argentina, Brazil, Chile, Mexico, Peru and Uruguay.Footnote 4

First, they described the heterogeneity of class structures, expressed in the contrast between Peru, the most predominantly agricultural country, and the other countries that had greater urbanisation and more widespread industry. Second, a common feature of all countries was the limited expansion of service classes, non-routine manual and highly skilled manual workers; and consequently a greater presence of classes of unskilled manual and agricultural labourers. These aspects defined the limited number of opportunities offered by class structures for upward mobility. Third, the authors found high rates of absolute mobility (similar even to those of European countries). Fourth, with respect to relative mobility, they concluded that: (a) the general levels of social fluidity did not differ significantly from those observed in early industrialisation countries; (b) Argentina, Chile and Mexico were more rigid with greater association between origins and destinations, but Brazil and Peru had higher levels of social fluidity; and (c) Latin America is characterised by a hierarchical pattern of social mobility, with greater distance between classes and polarisation of social mobility. In Europe, these aspects are dissimilar and more gradual.

Therefore, a balance between similarities and differences must necessarily take into account how polarisation is especially powerful in Latin American social stratification systems, and generates greater hierarchical distances between classes.

Solís and Boado, and the other authors in their national chapters, explored several models for associating class of origin and destination, for each country and comparatively between all of them, with an acceptable fit for the components of the core model in several Latin American countries, and for the fluidity and uniform differences models. But they eventually decided to adjust an RCII model modified by country. This model included the unequal distances between social class of origin and destination, while also considering the influence of class inheritance. This model postulates uneven boundaries between social classes, in origins and destinations, that modulate the possible movements, while also upholding the unequivocal importance of class reproduction.

3 Definitions, Data and Methodology

3.1 Definition of Social Classes

For all countries we agreed to apply the Erikson, Goldthorpe and Portocarrero (EGP) class scheme. For the Latin American countries some modifications were included as suggested and previously used by Solís and Boado (2016). The main modification consists of reclassifying self-employed workers without employees with low-grade occupations from class IVb to class VIIa. This change is made because, in the labour markets of Latin American countries, many of these workers declare themselves as independent, but are really subject to subordinate labour relations, in which they sell their work to one or more employers at the same time in highly precarious and unstable labour conditions (Solís and Boado 2016). For European countries, the classic proposal by Ganzeboom et al. (1992) was followed, which in turn was viewed as the most comparable with the data worked on before for the Latin American countries presented in this study. The scheme used is presented in Table 7.1.

Table 7.1 EGP scheme in seven classes used in the comparative analysis

3.2 Data

For Latin America, data was used as approved by several Latin American colleagues who contributed to the study by Solís and Boado (2016). For Europe, we worked with the data from the European Social Survey (ESS) for all countries except Italy, because the variables required to construct the EGP category were not available. Ganzeboon et al. (1992) specifications were followed to standardise classes using ISCO-88. For Italy, we used the Survey of Living Conditions (Indagine sul Reddito e le Condizioni di Vita) produced by the National Institute of Statistics and the classification was constructed following the same criteria as the ESS, making the necessary adaptations for such purposes in ISCO-08.

Similar and close age groups were selected for the European and Latin America surveys to ensure that the samples were relatively similar. Although an effort was made to standardise the data, there is no question that this primary comparative analysis could certainly be improved upon. Appendix 7.1 presents the characteristics of the surveys used for each country and the main methods for processing the information on mobility. The Table 7.2 presents the sample by sex.

Table 7.2 Sample used

Note that in order to avoid the influence of different sample sizes, in some operations the samples had to be balanced to the same size for all countries.

3.3 Models and Techniques

Since eighties Goldthorpe introduced the distinction between absolute mobility and relative social mobility, and subsequently the use of these meanings became widespread in the analysis of mobility using tables in which individuals are cross-classified by class of origin and of destination.

Absolute mobility brought together all the ways to measure social mobility and inheritance by relating cells in the aforesaid table, based on proportions (joint and conditional probabilities) and marginal dissimilarity indexes, among others. Relative mobility brings together all the procedures for estimating parameters that indicated the association between origins and destinations. Such is the case of the log-linear models and considerable developments that the scientific community introduced later, mainly associated to Research Committee 28 of the International Sociological Association.

Our lines of progress shall therefore be in two directions, towards an examination of the volume and characteristics of general social mobility, in order to contrast the rigidity or fluidity between Latin American and European countries; and towards the contrast of the differential effects of social classes on reproduction and social mobility.

We begin with the recognised results on the constant fluidity in Latin American countries as presented by Solís, Boado et al., and we draw on elements from European countries as proposed by Breen (2004), Vallet (2015) and Gil et al. (2017), to standardise the use of powerful models with proven goodness of fit, such as constant association models (which support the hypothesis of constant fluidity) and those of uniform variations (which support the hypothesis of the tendency to intergenerational fluidity or rigidity), to proceed by contrasting the communality between countries in terms of the volume of mobility.

In turn, we shall examine a model that emphasises class inequality, and that we believe, based on the experience of Solís and Boado (2016), to more clearly capture inequality between classes and its effects on social mobility than the core model, namely Row Column II (RCII).

3.3.1 Absolute Mobility

The analysis of social mobility considers a transition matrix that reflects both the forces of global expansion and contraction of certain classes and the propensity to inheritance and mobility between them (Hout 1983; Erikson and Goldthorpe 1992; Breen 2004; Fachelli and López-Roldán 2013; Solis and Boado 2016). Absolute mobility can account both for people who are better situated in the hierarchy or who have improved with respect to their origin (upward mobility) and for those who are lower down in the hierarchy than their parents or who have not yet achieved such a status (downward mobility). Reproduction, inheritance or immobility identify parents and children with the same social status, due to a transfer of occupation from parents to children or simply being in the same position due to a transitory situation. Absolute mobility is understood to mean the mobility observed directly via the frequencies in the table.

3.3.2 Relative Mobility: Rigidity and Fluidity

Analyses of relative mobility are concerned with something else. Greater attention is paid to the stability or variation that can be observed as a consequence of the association between origins and destinations. It is a more accurate way of responding to the concern manifested by Glass (1954) regarding social mobility and its plausibility. The authors have put every effort into overcoming the weakness of perfect mobility to generate hypotheses and models that capture the effects of inheritance and opportunities, and thereby present a more plausible explanation. Two perspectives of such models shall be applied here. On the one hand, those that make it possible to capture the fluidity or rigidity of societies. And, on the other hand, those that compare the distances and boundaries between classes and their effects on reproduction and mobility between classes.

The fluidity or rigidity of societies shall be measured by contrasting the social mobility between social origin and class of destination experienced by the generations in the sample. To do this, two hypotheses are used: one that supports constant fluidity and another that supports an increase in fluidity. In an open society, there will be greater possibilities for social mobility (more fluidity), while in a society where there is a large amount of social reproduction, and class origins (parents) have a major influence on children’s destinations, there will be a greater predominance of social rigidity (fewer movements and high inheritance). Hence, the odds ratios would indicate independence, non-association or non-influence of origin when their value is 1, any increase on 1, or any decrease (between 1 and 0) would indicate association. Carabaña (1999) calls this kind of mobility “doubly relative” as it is based on a measure that takes into account a dual relationship: a child who is in a category that comes from a certain class, in relation to another child that comes from a certain class that is taken as a reference.

The hypothesis of a society with stable mobility and stable reproduction is, as mentioned earlier, known as a constant social fluidity model (CnSF). In their details, relative mobility rates between origins and destinations remain constant across different cohorts. It is a homogeneous association model, with no interaction between the three variables, which implies that the relationship between origin and destination remains constant for each cohort or country in our case (Fachelli and López-Roldán 2015).

Another statistical model that is very widely used to compare between countries and work out how a pattern of association works is the Log-Multiplicative Layer Effect Model, better known as the “Unidiff” model (Erikson and Goldthorpe 1992; Xie 1992). This model constrains the variation between two or more mobility tables, for example, based on cohorts or sex, a pattern of common association and a changing term between the said tables. This parameter identifies the variations in the overall intensity of the association between class origins and destinations. The Unidiff model is useful for identifying the contexts in which there is greater or lesser social fluidity, given a previously specified fixed pattern of association (Solís and Boado 2016).

3.3.3 Relative Mobility: Distances Between Social Classes

We shall apply a set of models that have major discriminatory power to measure the distance between social classes, and which are more suited to comparative mobility research (Ganzeboom et al. 1991). These are known as statistical association, or Row-Column models, as presented by Goodman (1979), more specifically known as “RCII models”, developed after the Row-or-Column models (which were the first of the kind).

These are log-multiplicative models that measure the association between class of origin and class of destination, empirically estimating a ranking for the two (according to the association between them), thus generating a single parameter to measure the association between them. According to Torche and Wormald (2004) this procedure “reorders” the classes by estimating the distance between them and presenting a coefficient that measures the association between them. So, according to the aforesaid authors, there is no assumption a priori that the classes are ordered against one another, but that the order results from the empirical information about their association. For example, the less mobility there is between two classes, the more distant they are from each other in the ranking. They are also parsimonious models because they use a single parameter to express the association between social classes of origin and destination once ordered, so the results are easily interpretable. The RCII model serves as an empirical test of the ranking of classes, as it can be used to compare the empirically obtained ranking with rankings based on other criteria (income, education, etc.). So, to analyse whether the origin-destination association has changed between countries, the parameter that measures the association is allowed to vary freely between them, while the ranking of classes is assumed to be constant across the different countries. This makes it possible to assess whether the association remains constant, increases or decreases between countries.

RCII analysis can also be used to separate analyses of the cells of a transition matrix into those that are mobile and those that reproduce class, thereby obtaining association parameters for the cells that exchange individuals and association parameters for inheritance itself (measuring the effect of the diagonal). According to Torche and Wormald (2004) mobility studies have revealed a disproportionate tendency for people to remain in the class of origin (a phenomenon known as “class inheritance”) so it is useful to control for this by adding specific parameters for the main diagonal, which model the tendency to inheritance in each class. Otherwise, the discovered association may be completely due to class inheritance.

It seems appropriate to calculate the RCII for two purposes. On the one hand, to apply it to each country, in order to observe the distances between classes and draw conclusions at the country level; and, on the other hand, in the comparative context between countries, in order to observe the distances between classes, considering a level of scale that will allow us to observe how difficult it is to reach the highest class when coming from the lowest level, taking into account all the countries that appear in the analysis.

The model we use to analyse the distance between classes in each country (which we could call intra-country inequality) only uses two variables and analyses the intensity of the origin-destination association as fixed individually for each country and that is expressed in as many intrinsic association parameters (Phi, Ф) as there are countries, using a log-multiplicative procedure. To make the model parsimonious, the same scores are kept for parents and children (i = j), and the main diagonal is blocked (to reproduce inheritance exactly for each social class). The algebraic formula is presented below, where O = class of origin, D = class of destination, Phi (Ф) is the intensity of association, u and v are the scores for parent and child constrained to one dimension and delta (δ) is the main diagonal.Footnote 5

$$ \log\;{\mathrm{F}}_{\mathrm{i}\mathrm{j}\mathrm{k}}={\lambda}_0+{\lambda_{\mathrm{i}}}^{\mathrm{O}}+{\lambda_{\mathrm{j}}}^{\mathrm{D}}+\varPhi \left({u}_{\mathrm{i}}{v}_{\mathrm{j}}\right)+{\delta_{\mathrm{i}\mathrm{j}}}^{\mathrm{O}\mathrm{D}} $$

The model that we use to observe the differences between countries (which we can call inter-country inequality) analyses the intensity of the origin-destination association that is expressed in as many Phi (Ф) parameters as there are countries, using a simple heterogeneous log-multiplicative procedure that maintains the same scores between parents and children (i = j) for all countries except for the diagonal; i.e. the level of inheritance per country is allowed to vary (the main diagonal is blocked to exactly reproduce the inheritance for each social class). The algebraic formula is presented below, being O = class of origin, D = class of destination, P = Country, Phi (Ф) is the intensity of the association, u and v are the scores for parent and child constrained to one dimension for each country and delta (δ) is the main diagonal also by country in comparative terms.Footnote 6

$$ \log\;{\mathrm{F}}_{\mathrm{i}\mathrm{j}\mathrm{k}}={\lambda}_0+{\lambda_{\mathrm{i}}}^{\mathrm{O}}+{\lambda_{\mathrm{j}}}^{\mathrm{D}}+{\lambda_{\mathrm{k}}}^{\mathrm{P}}+{\lambda_{\mathrm{i}\mathrm{k}}}^{\mathrm{P}\mathrm{O}}+{\lambda_{\mathrm{j}\mathrm{k}}}^{\mathrm{P}\mathrm{D}}+{\varPhi}_{\mathrm{P}}\left({u}_{\mathrm{i}}{v}_{\mathrm{j}}\right)+{\delta_{\mathrm{i}\mathrm{j}}}^{\mathrm{O}\mathrm{D}} $$

Using this later model, we shall observe class inheritance comparatively by analysing the scores provided by the diagonal.

4 Results

4.1 Absolute Mobility

First, we will analyse the change in socio-occupational structure observed between parents and children in each of the countries by considering the seven classes defined in this study. We shall then analyse the upward and downward vertical social mobility in order to make an analysis of the subsets of countries and their main recent trends.

Structural change is a contrast between the proportions of the class statuses of the interviewees (children) and the statuses of their classes of origin (parent). First, in Fig. 7.1, we note that the changes were more similar in European countries, and were more disparate in Latin America. Then we observe a general increase in service classes in both continents, although it is bigger in Europe than in Latin America. Then there is the downward trend in agricultural classes due to the transformation of rural activities. In Latin America, Brazil experienced a particularly marked shift over a short period of time in the number of day labourers who migrated from the countryside to the city, an issue studied in detail by Costa Ribeiro (2012, 2014). This contrasts with several European countries in which the biggest change was instead in the decrease in skilled manual workers and also in those with lower grade skills. And finally, it can be seen that there is a decrease in smallholders in all countries except Brazil, while there are barely any changes to the petite bourgeoisie (IVab), the biggest being the increase for children with respect to parents in Brazil.

Fig. 7.1
A grouped bar graph of percentages versus Europe and Latin American countries compares the structural change between parents and children in seven classes. Class 7 b reports a negative 30% in Brazil while the bar for classes 1 and 2 reports 25% in Finland.

Structural change between parents and children. Source: The authors

Another form of global and comparative analysis is to consider absolute mobility in terms of vertical mobility, examining the increases and decreases. We group the seven classes from the original EGP scheme into four macro-classes: the first only considers the service class (I + II), the second consists of the routine non-manual class up to skilled and semi-skilled manual employees (IIIa+b, IVa + b y V + VI), the third includes only non-agricultural lower grade manual workers (VIIa) and the fourth contains the agricultural classes (IVc and VIIb).Footnote 7

We first note the greater presence of reproduction or immobility among males in Latin American countries than in Europe. In turn, there is a predominance of vertical upward mobility in the latter, as opposed to reproduction or downward mobility. Meanwhile, two trends are observed in Latin American countries: those where reproduction is more widespread than upward mobility (Chile, Argentina and Uruguay), and those where upward mobility is the main feature (Brazil and Mexico).

A remarkable characteristic is the mix between European and Latin American countries when ordered by upward mobility (Fig. 7.2). Reflecting this pattern, there are not many differences between absolute rates, except for those mentioned earlier for Uruguay, Argentina and Chile, while Finland is at the opposite extreme (high mobility and less inheritance than in the other countries) in the case of males.

Fig. 7.2
Two graphs of percentages versus countries plot immobility and descending mobility in bars and an increasing trend line for ascending for men and women. Immobility is high in all countries for both genders.

Vertical Mobility. Source: The authors

However, the pattern for women in Latin America and Europe is different. Vertical upward mobility is greater than reproduction in all Latin American countries except Uruguay. Immobility and upward mobility have similar values, except in Great Britain where immobility affects a significant percentage of women.

4.2 Relative Mobility: Rigidity and Fluidity

Analysis of relative mobility, as mentioned earlier, reports possible changes between origin and destination when we isolate the effect of marginals, i.e. when changes in absolute mobility are not considered. We want to work out the extent to which the opportunities for children to attain a certain position are conditioned by their parental origin, and whether this pattern of behaviour varies or not over time, and also between countries, and also to find out which countries have the greatest degree of social fluidity and which are more rigid.

Table 7.3 presents the results of applying the constant fluidity and Unidiff models to compare and contrast all the studied countries. Argentina is taken as the baseline for application of the Unidiff model. First, constant or common fluidity is preferable to Unidiff in the case of males (higher negative BIC value). However, the L2 values, both for men and women, show that there are significant differences,Footnote 8 so we choose the Unidiff model and thereby interpret the differences obtained.Footnote 9

Table 7.3 Trends in social fluidity. Origin-destination by country

The results show that European countries are more fluid than Argentina (the country taken as our reference) and than Latin American countries in general. This trend is particularly noticeable among females, as clearly shown in Table 7.3 and Fig. 7.3.

Fig. 7.3
A grouped bar graph compares social fluidity values for men, and women, along with a total and lineal total for 10 countries. The reference line is at 1 which is the same as the value of Argentina. Chile and Mexico have crossed the reference line. The lineal total plots a decreasing trend.

Social fluidity. Unidiff values by country. Source: The authors

Brazil is an exception among Latin American countries, especially as a result of its massive de-ruralisation. Uruguay is somewhat more fluid than Argentina in terms of men and women. Chile is not fluid for either sex, and Mexico is more so among women than among men.

Meanwhile, European countries have greater fluidity, and clearly more so among women, an aspect that is particularly prominent in Great Britain and Finland.

4.3 Social Distances

4.3.1 Relative Mobility: Distances Between Intra-Country RCII Classes

To capture in detail what lies behind the volume of change, the Row-Column (RCII) models can first be used to look at each country. Figure 7.4 shows that all countries experienced mobility processes (size and sign of the bars) and changes in the intensity of the association (size of the Phi shown beneath the horizontal axis).

Fig. 7.4
Two bar graphs compare the scores of men and women from 7 classes in different countries, followed by the phi values. The bars for classes 1 and 2 have the highest negative values in both plots and the highest positive values for class 7 b.

Distance between intra-country classes (scores). Source: The authors

Hence, the distances between classes in terms of RCII score are indicative of how frequent relative mobility is between them. These RCII model scores can be used to compare hierarchical order between social classes in each country.

In the Fig. 7.4, that distance is represented by the gap between the vertical bars that separate each of the classes, so when two classes appear very close to each other, this tells us that there is greater social fluidity between them, while a wider gap tells us that there is less fluidity, so we can assume there to be a greater social distance (Solís and Boado 2016).

The obtained results show that all the analysed countries share the same hierarchical order in their class structure. In this order, the agricultural classes at one extreme and the service and non-routine manual classes are at the other, and the existence of variations between countries in terms of hierarchical distances is observed as in Solís and Boado (2016).

In Latin American countries, there is eloquent polarisation of classes among men, whereby social extremes are more distant, and intermediate social classes are closer together. However, the association coefficients reveal Brazil and Chile to be more unequal than the other countries, while in Argentina, Mexico and Uruguay the extremes are less pronounced. The same characteristics are observed for Argentinian and Uruguayan women. This is not the case for other countries, where the distances are smaller or larger, indicating less verticality and hierarchical sequence.

In Europe the distances between the extreme classes are certainly also wide, but less so in the cases of men in France and Finland. Spain has the highest Phi coefficient. The class difference between women is similar to that for men, but they have greater coefficients of association than males in Spain and Italy and smaller ones in France, Great Britain and Finland. Note that the female smallholders in Great Britain have a different pattern of association to women in other European countries.

It is important to note that Latin American countries have a higher value than European ones for the Phi parameter, i.e. the general association between classes. This expresses greater differences in the proportions of inequality in the former. These differences will be taken into account as we scale classes in consideration of all countries at the same time in the following section.

4.3.2 Distances Between Inter-Country RCII Classes

By taking the parameters of association and observing the inter-class differences between countries, we now take a comparative look at our results (Fig. 7.5).

Fig. 7.5
Two scatterplots of log odds ratio versus countries plot a decreasing trend for 7 classes of men and women. Brazil has the highest value and Finland has the lowest. The segments for each country have values in boxes.

Distance between inter-country classes. Access by day labourers to the other classes. Note: The values in boxes are the Odds Ratios of attaining classes I + II (The odds ratio graph is shown in the Appendix). Source: The authors

First, we note that the hierarchical order between social classes spatially represents an order that we commonly use for classification, and which justifies why researchers frequently ‘downgrade’ class IVc, whereby smallholders are allocated a status associated with the lower classes (VIIb and VIIa).

RCII model scores transformed into log-odd ratios show that the hierarchical order expresses four areas of proximity between classes: the highest two are very close, there is a certain distance between the intermediate ones but they are more or less together, IVc is somewhat isolated from the means and VIIb, which is the one taken as a reference.

In all the societies analysed, day labourers that are the children of day labourers are the ones that face the biggest barriers (or those that have to cover the most distance) if they want to achieve the position of children in the service classes that come from parents of the same class. But these distances are also big for smallholders.

Another clear aspect is that the RCII scores, by including each society’s association coefficients (Ф) in comparative terms, show us that relative mobility between classes is much more frequent in European countries than in Latin American ones. And they particularly reveal that in societies such as Brazil, Chile and Argentina, mobility barriers are high for both men and women.

In the case of European women, the situation is similar to that of men except in the case of Great Britain where the greatest inequality is expressed among men and not so much among women, who face fewer barriers and where the classes are closer to each other. Also note that Spanish women face major barriers between classes, which is more similar to Latin American behaviour than European.

4.3.3 Distances Between RCII Classes: Inheritance

On the other hand, we propose analysis of such a relevant aspect as inheritance, in order to show which classes are the most and least reproductive. Since RCII models can be used to analyse the main diagonal separately (i.e. to reproduce the inheritance exactly for each social class), we now present an analysis of the scores obtained for the diagonal of all classes and all countries, comparatively valued.

The following graphs show the propensity to reproduce between classes (Fig. 7.6). Zero is the boundary between the classes that reproduce the most (highest positive values) and the least (which have negative values).

Fig. 7.6
Two scatterplots of values versus countries plot cross points for 7 classes of men and women. Some of the plotted points are grouped. Class 4 c has the highest values in both plots.

Propensity of reproduction within each class, taking into account the distance between inter-country classes. Source: The authors

First, we observe major regularity between countries, because the class that is most widely reproduced, i.e. is most intergenerational, is IVc in all countries in the case of men. Meanwhile, the classes with the lowest chance of reproduction are VIIb and IIIab in all countries.

We should particularly highlight that although Brazil was more fluid when the diagonal was not blocked (in relative mobility), when observing here the effects of reproduction by class (diagonal by class), IVc is the least rigid in comparison with the other countries. We also note that together with Italy, it has more mobile VIIb classes than in other countries.

In the case of women, class IVc is also the most reproductive (and extremely so in Mexico). However, this is less the case for Brazilian and Finnish women, due to different circumstances, and they are exceeded in terms of reproduction by other classes, namely VIIa, V + VI and I + II in Brazil and I + II and VIIa in Finland, which are practically at the same level as IVc. For its part, the least reproductive class is VIIb in all countries (and particularly so in Uruguay).

These results reflect how it is from agricultural areas (day labourers in particular) that much of the mobility has come in Latin America and in Europe, especially if there is a lack of assets or means to develop farms. Meanwhile, among urban classes the propensity for social reproduction is more equidistant among other classes, although the service, small business owner and skilled worker classes are still observed to be preponderant.

5 Summary and Conclusions

The general purpose of this chapter has been to make a contribution to the comparative analysis of social inequalities based on an analysis of intergenerational social mobility among the countries of the INCASI network.

We use the classical analysis methodology for such studies by addressing absolute mobility and relative mobility, investigating social fluidity and rigidity and using models to measure the distance between social classes and that are most suitable for comparative research.

We use data that has been approved previously by Latin American researchers and did likewise with European sources, trying to make the interpretation of the different surveys as comparable as possible.

The results obtained are summarised below, after which the section ends with a discussion of their implications for the analysis of inequality.

With respect to the analyses of absolute mobility, we have been able to show that structural change has been very important both in Latin American and European countries. We note greater similarity in class movements in Europe while in Latin America these changes are more abrupt. The sharp decline in agricultural classes in these countries together with a sharp decline in industrial classes in Europe showcase two different stages of the industrialisation process, as observed by Ishida and Miwa, which can be summarised as high rates of de-ruralisation, these being greater among day labourers in late industrialised countries and a more pronounced process of deindustrialisation in European countries. The growth in service classes shows how pronounced this post-industrial stage is in Europe, but this process is also visible in all Latin American countries, albeit less pronounced and with more nuances than in Europe.

When analysing absolute social mobility in a more compact manner by converting the seven EGP classes into four macro-classes, we observe the importance of immobility, a relevant issue in this field of study as its volume is indicative of the strength or inertia of social reproduction. More than 40% of the people in the group of countries analysed are in this situation, with there being just over one percentage point more in Latin American countries than in Europe.

This factor is less important if a labourer’s child is a labourer or if the child of a day worker is a day worker. The situation becomes more important when considering the aspects of social inequality that this entails, i.e. high reproduction rates in proprietary and service classes, for as the sociological literature has extensively demonstrated the inheritance of property, knowledge, client portfolios, social networks and opportunities of all kinds are an important factor for the perpetuation of inequality processes and the generation of conditions at institutional or meso-social levels that generate path-dependent effects that are difficult to reverse. So, the closing off of classes creates an additional obstacle for movement by other classes towards the higher levels.

However, by lowering this global 40% immobility to each of the countries, we have found that it is men in Uruguay, Chile and Argentina who have the highest levels of reproduction. Meanwhile, it is women who have experienced the greatest upward mobility (43.6%). Finally, the level of global downward mobility is around 17%.

Leaving aside structure, the marginals in the table, to start assessing relative mobility, we observe that European countries are more fluid and Latin American countries are more rigid, except Brazil, as commented earlier given that class IVc is the least reproductive compared with the other countries. In turn, Chile, Mexico and Argentina stand out as the most rigid when compared to the rest and France, Italy, Great Britain and Finland are the most fluid. The fluidity of these countries is driven to a greater extent by women.

When studying the differences between classes within each country, but not comparatively, we observe a shared hierarchical order, with the agricultural classes at one extreme and the service classes at the other, which shows the polarisation between classes as well as various intermediate situations depending on the particularities of each country. The Phi coefficients for each country show that men in Brazil and Chile, as well as women in Mexico and Chile, have the greatest relative inequalities. On the other hand, men from Italy, Spain and Great Britain and women from France, Great Britain and Finland have the fewest differences.

To put all the countries in relation to each other, we apply a model to analyse them as a whole and obtain comparative relative inequality measures between classes and between countries. We thereby observe that relative mobility between classes is much less frequent in Latin American countries than in European countries. And we could especially observe that mobility barriers are very high in Brazil, Chile and Argentina, for men and even more so for women, compared to the other countries.

Finally, by analysing class inheritance separately, we were able to corroborate a uniform pattern of behaviour in all countries, observing that the smallholder class is the most reproductive, and that this is the case in all countries analysed. For its part, the class that contributes the least to reproduction is the lowest, that of agricultural labourers.

Before going back to our hypothesis, we would like to focus in on the importance and interest of the analysis of social mobility for the study of social inequality. We could ask what relevant information we can get out of social mobility analysis to serve the purposes set out for the INCASI project, i.e. to address the issue of social inequality. The answer is quite a lot. Analysis of social mobility allows us to briefly analyse inequality between social positions, by classifying people into different classes that are approximated by different socio-occupational groups (basically through studies of absolute mobility), and by analysing the inequality of access to these social positions (which we do through studies of relative mobility), which in themselves are already unequal. What truly matters here is that both approaches allow us to observe something ‘invisible’, namely social inequalities (distances) between social classes. We are getting to the heart of sociological analysis by analysing social stratification.

Our hypothesis, as mentioned earlier, was based on a classical idea drawn from the comparative literature on social mobility, which Ishida and Miwa (2011) put to the test by including countries that are not usually analysed in such studies: those of late industrialization. The authors proposed that there is a relationship between social inequality and the distribution of opportunities, and that this is grounded on the fact that societies that joined the industrialization process late generated greater social inequalities and also more disparate concentrations of opportunities for social promotion, which is expressed as less social openness.

In this study, we believe that we have provided evidence in that regard, using the three types of countries analysed by the aforesaid authors, namely Great Britain and France, which can be classed as early industrialised, those of late-late industrialisation that we might find in Latin American, and finally intermediate early-late industrialisation countries that might include Spain, Italy and Finland.

Based on the analyses of social fluidity that account for the degree of social openness on the one hand and, taking into account the analyses with RCII models that provide information about inequality between classes and between countries on the other, we can only corroborate the authors’ hypothesis that “the later and faster industrial development is, the higher the inequality and the lower the social fluidity” (Ishida and Miwa 2011: 9–10).

6 Discussion

We now present a discussion both of the contributions that we consider this study to make to social mobility, and of its limitations, as well as making some suggestions for the INCASI network’s future research programme.

The study that we have presented in this chapter can be set in the context of the tradition of so-called fourth Generation social mobility studies, which are, according to Treiman and Ganzeboom (2000), those that were made in comparative terms, such as those that addressed the issues of education (Shavit and Blossfeld), the transition from school to work (Shavit and Müller), the achievement of status (Ganzeboon and Treiman), class structure and class awareness (Wright), economic attitudes (Kelly) and the study of social stratification in Eastern Europe (Szelényi and Treiman). We could add to this brief summary the ground-breaking contribution made by Ishida and Miwa that included late industrialised countries to traditional analyses of European countries and, above all, the United States.

In turn, this chapter also covers the tradition of, and makes wide use of its contributions, third generation mobility studies, led by Erikson and Goldthorpe who proposed the use of discrete variables and introduced a major development to the relevant methodologies (loglinear analysis) by considering classes to be intrinsically discrete and not necessarily ordered hierarchically. This is also the generation that introduced the flexible models developed by Goodman, such as the RCII that we have used extensively.

Although we do not apply the methodology of second generation studies, we do include some of the more theoretical contributions highlighted by Ganzeboom et al. (1991) such as the studies by Blau and Duncan (1967) and all those that deal with the scales of occupational prestige (Treiman 1977), which primarily use continuous quantitative variables and provide evidence to support modernisation theory. Finally, the first Generation is represented by such pioneering studies as those by Glass (1954) and Sorokin (1959), which albeit more rudimentary, using only two or three class groups, raised concerns that continued to feature in the biggest research questions of subsequent generations, such as the interest in the extent to which countries differ in terms of their mobility patterns (Lipset and Zetterberg 1959), whether the highest mobility rates occur in industrialised societies and the interest in analysing the relationship between the degree of political stability and mobility (Fox and Miller 1965), and what mobility processes were like under socialist regimes (Connor 1979).

In a way, we are heirs of this broad tradition and our hypothesis is well in line with the concerns that were raised by the first generation of studies but that have been a common thread throughout all the generations. Greater and better quality data is now available, so it is time to re-examine those foundational concerns. This leads us on to the main limitations of this study.

The study has limitations in terms of its scope, as it only analyses the relationship between origin and destination, it is fairly descriptive in nature and does not look in depth at the substantive processes that explain the analysed changes. Such a profound analysis of each of the countries is beyond the scope of this version. A lot could be done to improve the comparability between surveys both in terms of sample size and standard definitions of classes. We have worked with the data that was standardised for Latin America by Solís and Boado and colleagues from each Latin American country who worked so painstakingly with their national surveys, which enabled us to propose a preliminary comparison between Latin America and Europe that is limited to INCASI countries. However, the differences to the findings by Solís and Boado (2016) need to be taken into account, such as those in terms of the historical periods analysed. For example, those authors used data from 11 countries analysed by Breen from the 1970s to 1990s, while our study considers five European countries. Likewise, the Latin American countries are not the same either, as those authors include Peru, a country that adds considerable fluidity to the group of countries analysed, but their comparisons do not include Uruguay, a country that does appear in our sample and for which a national sample has been produced for the first time. Finally, the methodology that we used is also improvable and expandable. There are several other models that could be applied but that we have not presented in this chapter. Our findings should be taken as a starting point for the comparative research project that we intend to conduct in the future, and which shall begin with review, improvement and extension of these findings.

We therefore proceed with some proposals for the network’s future comparative research project. Briefly, we propose two aspects, one methodological and the other substantive. In terms of methodology, the first step would involve looking further into the effect of education on social mobility by including one more variable. The goal is modest but the methodological development is new and innovative, and could be deemed the fifth Generation of mobility studies following the work by Breen (2010) using counterfactual models applied to the breakdown of the OED (origin, education and destination) triangle. These models require large, good quality samples, and very recent efforts have been made to produce just that in the case of Latin America. Second, in theoretical terms we need to address institutional issues, which are extremely relevant for comparative studies, enter into such pending debates as substantive issues related to unequal conditions and results, and also incorporate as many aspects as possible of the Analytical Model on Social Inequalities and Trajectories (AMOSIT) proposed by the INCASI project in order to relate not only macro-social aspects but the whole spectrum of elements that condition situations of inequality. Although we are aware that this is highly ambitious, and also very demanding in terms of data, we are convinced that this is the direction that we need to move in.