INTRODUCTION

The focus of this paper is to make the first steps of a comparative analysis of time variations in intergenerational social (class) mobility over birth cohorts in Argentina and Spain¹. Intergenerational mobility will be analyzed according to positions in the class structure, taking into account those of the interviewees and their fathers, based on national surveys.

As Breen (2004, pp. 4-5) –among many others- recalls, two main hypotheses on the expectations of variability in patterns of social mobility have been proposed in the literature. On the one hand, the early work by Lipset and Zetterberg (1959) proposed the hypothesis of a similar overall pattern of observed social mobility in industrial nations (LZ hypothesis). In contrast, Featherman, Jones, and Hauser -FJH- (1975) proposed that there are similarities in variations in intergenerational social mobility -in western industrial societies- and that these changes are linked to historical and cultural patterns, net of structural variations - “cross-temporal stability and a cross-national similarity”-, as Ishida and Miwa (2011) stress. When this hypothesis was revised, it was pointed out “that mobility chances are invariant once variations in origin and destination distributions have been controlled” (Grusky & Hauser, 2001, p. 336). Furthermore, Breen (2004, p. 4) notes that the high variability of marginal class distributions in mobility tables was a limitation for that hypothesis based on absolute mobility. In their exercise comparing “convergence and divergence” in 16 countries, Grusky and Hauser (2001) found strong support for the FJH hypothesis, adding that “uniformity in mobility regimes is not limited to highly industrialized countries but may extend across levels of economic development” (Grusky & Hauser, 2001, p. 341). Furthermore, Erikson and Goldthorpe (1992) observed a “basic similarity” in such relative mobility, and also supported the common basis of mobility proposed by the FJH hypothesis. Similarly, Breen (2004), after finding a similarity in patterns of origin-destination class association between countries, signaled differences in the strength of association (or fluidity). However, if periods and/or cohorts are taken into account, Breen and Luijkx (2004) question the convergence thesis of social fluidity, given that they found contradictory results for Great Britain and Germany (variations over cohorts only in Germany). On the other hand, Ishida and Miwa (2011), who distinguished within “late development countries” those “early late” (which industrialized at the same time or later than Japan) and “late-late industrializing nations” (which reached industrialization after Japan), found similar patterns of observed absolute mobility, but with “substantial variations in the level of social fluidity”, adding that “we cannot explain these variations satisfactorily by referring to the timing of industrialization …” (Ishida & Miwa, 2011, p. 19). In addition, for Latin American countries –most of which might be considered “late-late developers”-, Torche (2014) finds that even though they tend to exhibit some weaker income mobility, class mobility looks similar to that of industrialized nations. In a benchmark study that compared several Latin American countries (Solís & Boado, 2016) the authors conclude that there are differences between the structures of Latin American countries (based on the greater weight of the agricultural classes and on the smaller expansion of the service and non-routine manual classes). However, when they compare these countries with European ones, they find certain common features, such as the similarity in the high absolute rates of intergenerational social mobility with respect to absolute mobility and also similar general levels of social fluidity.

On the basis of previous discussions of convergence or divergence, we will analyze whether Argentina and Spain are in any way closer to either of these patterns.

JUSTIFICATION OF THE STUDY, LITERATURE REVIEW AND HYPOTHESES

In social mobility literature, particularly that on relative mobility, stability, decline or trendless fluctuations have been observed in the association between class origin and class destination. For the trendless fluctuation hypothesis, see among many others Erikson and Goldthorpe (1992), Bukodi, Goldthorpe, Waller, and Kuha (2015), Hertel (2017), and Breen and Luijkx (2007) for a discrepant view. For Latin-America, Solís and Boado (2016) support Breen and Luijkx’s point of view to a greater extent when they analyze the social fluidity of five Latin-American countries in comparison with eleven European ones, but support a trendless fluctuation when comparing five Latin-American countries with each other. This is still a prevalent discussion in the field. And two approaches might be linked to it: cross-national comparisons between countries and evolution over time within specific nations, or the combination of both. Breen and Luijkx (2007) highlighted the importance of dealing “with differences between countries rather than change through time” (p. 102). They then moved on to study change through period and cohort analyses. Seemingly, Bukodi, Paskov, and Nolan (2017, p. 4) recognize the existence of these two “strands” but note that they will study the former. We will look at both approaches.

As a kind of general reference, the concluding remarks of a description of mobility processes from Chan (2009, pp. 47-48) will be taken into account:

i) that explorations have shifted their focus “from testing no trend in absolute rates (LZ hypothesis) to testing no trend in relative rates (FJH hypothesis)”;

ii) that there is a wide variation in absolute mobility given national historical and political conjectures, particularly the timing of industrialization; and

iii) that a similarity of relative mobility rates (social fluidity) across nations and over time was found.

Our countries of interest -Argentina and Spain-, which might be considered late industrializing nations, are not among the late industrializing countries in which social mobility has been previously explored (Ishida, 2008; Ishida & Miwa, 2011). It is not common to find class mobility comparisons of European and Latin American countries; this is a deficit we aim to start covering with this study. For example, in Ganzeboom, Luijkx, and Treiman (1989), only Brazil was taken into account. The studies by Erikson and Goldthorpe (1992) and Breen (2004) are devoted to European countries, and Spain was not among them. The only one that includes Spain is a recent study of 30 European nations (Bukodi et al., 2017). Solís and Boado (2016), as we already mentioned in the introduction, find interesting similarities as well as differences when they compare Latin American countries with the European ones analyzed by Breen and Luijkx (2004). Ishida and Miwa’s unpublished study (2011) might appear to be an exception (Brazil, Chile, and Mexico were included), but it is devoted to “late developing countries”, and no European nations are considered. So, we find here an additional interest for our proposed exploration.

In the context of our inquiry, we note that several recent mobility studies of European countries –particularly Great Britain, which is a common reference in mobility models- were worried about the comparative limitations of earlier data with respect to that of the present (among others, Bukodi et al. 2015). While these authors have questioned the adequacy of present data for mobility studies in Great Britain, we will note in passing that the evolution of Britain’s class structure and mobility showed an opposite tendency when we used their “new” rather than “old” data as a standard reference for our comparative international mobility models.

Based on the preceding considerations, two central questions will somehow guide our approach: 1) What have been the changes –if any– in absolute mobility rates in Argentina and Spain, when individuals’ class destinations are compared with those of origin? 2) What has been the comparative evolution over time of the strength of class association for Argentina and Spain? More specifically, what was the evolution within each country across four similar birth cohorts?

Linked to those central questions, two working hypotheses will be considered as guides:

1) Given intergenerational changes in labor movements from rural to urban environments, plus the growth of the service class –or “salariat”– we expect to find a high degree of absolute mobility in both countries; and 2) following international literature as well as previous research by the present authors, we would expect to find a stable net association –controlling for structural changes– of class origins and class destinations across birth cohorts in men, and social fluidity in the case of Spanish women.

ARGENTINA AND SPAIN IN THE CONTEXT OF DIFFERENT MOBILITY TRENDS

We must begin by noting that Spain and Argentina’s economic trajectories are quite different. Spain was, until the last third of the twentieth century, a generally poor country with high emigration, backward agriculture and some industrially developed poles (Catalonia, Basque Country). At the same time, Argentina was a country that attracted migrants, was much richer, and was based on modern, competitive agriculture, together with a moderate but dynamic process of industrialization in its urban centers. In the late 20th century, Spain began to develop, with considerable modernization of its entire productive structure, while Argentina -despite a major modernization process- did not manage to modernize its productive structure in the secondary sector, in which it barely competitive unlike it is in the agrarian sector). Something they share in common is that neither Spain nor Argentina could be considered a “developed” country, though they might be called “late industrializing countries”. Spain moved a bit closer to this label when it joined the European Union, but it has not yet reached the productivity levels of Europe’s biggest economic centers, or the USA, Japan, South Korea and Australia. Measurement of the pace of industrialization might be a relevant variable for comparisons between both countries, but this should be done with caution, given their similarities in this regard (Carreras & Tafunell, 2010; Tortella & Núñez, 2011; Míguez, 2005; Gerchunoff & Llach, 2010). However, the World Bank considers Spain a high-income country (with chronically high unemployment, we must add) and Argentina a middle-income country (with chronically high inflation rates, we must add). In spite of our caution on this point, the contrast between both being late industrializing countries (Argentina being late-late and Spain early-late if we use the terminology proposed by Ishida and Miwa, 2011) but with relevantly different income levels might add interest to our explorations.

With respect to the background of studies on class mobility in our countries, we note that in Argentina previous research was based on the Buenos Aires Metropolitan Area (Germani, 1962; Beccaria, 1978; Jorrat, 2000). The first national survey for Argentina was carried out in 2003 (Jorrat, 2005). Renewed interest in the country has been aroused by doctoral dissertations and publications devoted both to the analysis of absolute and relative mobility and using birth cohort analyses observing the prevalence of constant social fluidity in intergenerational class mobility (Dalle, 2015, 2016, 2018; Pla & Rodríguez, 2015; Quartulli, 2016; Jorrat & Benza, 2016; Jorrat 2016).

Studies of intergenerational mobility in Spain can be divided into two stages: the foundation stage and the current stage. In the foundation stage, the analysis of social mobility was started by two fundamental contributions: those of Echeverría (1999) and Carabaña (1999). In response to these early works, and in order to broaden the perspective of analysis to include women (traditionally the invisible half of many classic analyses of mobility), Salido (2001) reported on female occupational mobility in Spain. After an unproductive period with no new sources of data for analysis, studies of intergenerational mobility in Spain made a comeback in the late 2000s. The work by Marqués Perales and Herrera-Usagre (2010) updated the empirical research on intergenerational mobility with new data from the Instituto Nacional de Estadística (INE, 2008) 2005 Survey of Living Conditions (Encuesta de Condiciones de Vida, ECV). Their analysis of social fluidity provided new evidence that the constant fluidity model of European countries is well suited to the Spanish case. The same analysis with the 2005 and 2011 ECV was replicated by Fachelli and López-Roldán (2013. These studies evidenced the need to include both daughters and mothers, the integral model of dominance, in order to consider the invisible half of society. They also put forward the hypothesis of a constant trend of social fluidity over time (for individuals born between 1951 and 1985) but with non-uniform variation, with a trend towards greater global social openness as revealed by differentiated gender patterns: male rigidity is combined with greater female fluidity, generating an overall attenuated trend of social fluidity consistent with other results at the European level (). These interpretations are consistent with some of the conclusions reached by . With a more extensive dataset, Gil-Hernández, stated that fluidity could be observed among men too.

Before closing this point, it must be noted that in our present analysis for both countries temporal variability was based on data from the same four birth cohort (people born between 1951 and 1985), of around 8/9 years each: 1951-1959, 1960-1968, 1969-1976 and 1977-1985), a relatively “short period of time”². A scheme with some basic historical descriptions of our cohorts is presented in an Appendix (Table A1), just for contextual reference.

It is no easy task to associate socio-economic periods with results of social mobility. Too many uncontrolled variables would be involved in this endeavor. Difficulties arise for several reasons:

1) For early periods, only male mobility was studied, and the late direct incorporation of women into the class structure is well known.

2) Such an analysis would require several samples across different decades.

3) When absolute mobility rates are considered, the influence of structural changes on social mobility is particularly complex, depending above all on class origins (Carabaña, 1999).

4) It might be said that there is a kind of reciprocal interdependence between structure and mobility, so it is difficult to establish the extent to which social positions appear or disappear as a result of the demands of the economy or the availability of a workforce with different levels of training or education.

5) Finally, the most important difficulty might be an attempt to relate changes described by economic history over the considered time spans with changes that the mobility tables show for birth cohorts.³

For all these reasons, the analysis of social mobility in this study will not be specifically linked to the economic and political processes that we describe for each cohort or span of time in the Appendix, which are presented for merely illustrative purposes.

As we noted, Argentina and Spain share the common trait of not being early industrialized countries, with their own timeframes and magnitudes of late industrialization. Accepting differences between the two countries in socio-historical economic terms, this study aims to find out whether similarities or differences prevail for the trends in absolute and relative mobility, following the lead of equivalent international discussions.

DATA AND MODELS

Data sources

In the case of Spain, our explorations take into account data from the 2011 Survey of Living Conditions (INE, 2013), that corresponds to the European Union Statistics on Income and Living Conditions (EU-SILC), based on a module of intergenerational transmission of poverty that gathers, through retrospective questions, information about the socio-economic situation that adults aged 25-59 years old at the time of the survey—i.e. people born between 1951 and 1985—had experienced during their adolescence (at the age of about 14). There were 6948 males and 6740 females.

In the case of Argentina, eight surveys were carried out from 2003 to 2012.⁴ For the purposes of comparison with Spain, those born between 1951 and 1985 and persons aged 25-59 years were taken from these surveys, and cohorts were built with the same range of birth years as in the Spanish study, about 8-9 years each (1951-1959, 1960-1968, 1969-1976 and 1977-1985)⁵. There were 4171 males and 4084 females.

The Argentine studies are representative national surveys of people aged 18 and over, conducted by the Centro de Estudios de Opinión Pública (CEDOP) at the University of Buenos Aires. The surveys had a similar socio-demographic data structure, particularly the questions on origin and destination occupations. Different tests, plus distinct pooling of some of these surveys -very close in time- in other studies that found similar results, are additional support for our procedure.

The class categorizations –for Argentina and Spain- are based on Ganzeboom and Treiman’s algorithm for the CASMIN-EGP class scheme (2019). Given that some variations in the number and grouping of categories for comparative purposes are used, in each case the used categories will be specified.

Table 1. Seven EGP class schema

I+II	Service class	I) Higher-grade professionals, administrators and officials; managers in large industrial establishments; large proprietors. II) Lower-grade professionals, administrators and officials; higher-grade technicians; managers in small industrial establishments; supervisors of non-manual employees.
IIIab	Routine non-manual employees, a) higher and b) lower grade	a) Routine non-manual employees (higher grade) in administration and commerce; b) Sales and services (lower grade).
IVab	Non-rural “Petty bourgeoise”	a) Small proprietors, artisans, etc., with employees. (Employer) b) Small proprietors, artisans, etc., without employees. (Self-employed)
IVc	Farmers and smallholders, other self-employed workers in primary production	Employer or Self-employed.
V+VI	Lower-grade technicians; skilled workers	V) Lower-grade technicians; supervisors of manual workers; VI) Skilled manual workers.
VIIa	Semi- and unskilled manual workers	Semi- and unskilled manual workers (not in agriculture).
VIIb	Agricultural and other workers in primary production	Semi- and unskilled manual workers in agriculture (Agricultural laborers: agricultural and other workers in primary production).

Models

We propose to follow some predominant analytical trends that rely on traditional techniques for our explorations of intergenerational mobility.

As usual, these types of analyses consider a transition matrix between origins (usually the main occupation of the father during the respondent’s adolescence) and destinations (the respondent’s current or past occupation)6. Thus, absolute mobility refers to changes or persistence in the current positions of individuals in comparison with those of origin. Immobility (reproduction or inheritance) identifies the fact that parents and children share the same social class due to a transmission from parent to child or simply because they share that status in a transitory situation.

Relative mobility identifies trends in class movements regardless of changes in the marginal totals of the matrix. Relative mobility (or social fluidity) is interpreted as an opportunity to reach one destination rather than another depending on origin.

To explore this type of mobility, we use log-linear and log-multiplicative models (Xie, 1992; Powers & Xie, 2008; Wong, 2010). The independence model (or conditional independence model) is recognized as the hypothesis of “perfect mobility” and assumes that destinations (D) are not related to origins (O), thus implying the existence of statistical independence. This model does not fit any known real society but is considered a reference against which other models are compared, such as the constant association model (or full interaction constant social fluidity), which assumes that class origin-destination associations are constant over birth cohorts. That is, “the degree of social fluidity within the class structure is at a constant level” (Goldthorpe, 2016, p. 97).

Another common model, the uniform difference model (Unidiff), is a log-linear multiplicative layer effect model (Xie, 1992). Under the assumption of a stable structure in the association between origins and destinations, “this model is able to detect differences over cohorts in strength of association” (Vallet, 2006, p. 13). Units of analysis can also be countries, of course, as we will later show. The possible differences in association, measured by pertinent coefficients, indicate stronger O-D association for each cohort if those coefficients are above 1 and weaker association if they are below 1 (being 1 the value assigned to the reference category). Another version of this model is the linear uniform difference model, “where parameters are constrained to be linear over decades” –or over birth cohorts as in our case- (Breen, 2004, p. 51).

Finally, we offer results following the log-multiplicative regression type approach, proposed by Goodman and Hout (1998), a model that, as Vallet (2006) notes, is able to detect differences over cohorts in both pattern and strength of association (p. 18).

RESULTS

Gross changes in occupational class structures

In Table 2 below we show changes in the occupational structure of both countries, for a six EGP class scheme, by sex (Table 2a) and for both sexes (Table 2b). We will comment on the former.

If we compare the occupational classes for males resulting from the marginals of the tables of fathers and their sons and daughters in Argentina, we observe the change in structure reflecting an increase in male non-manual classes (I-II-III) of 10.4 percentage points (33.5 – 23.1) versus a 19.0 percentage point increase for males in Spain (40.7 – 21.7). This is more so for the top of the service class I in Spain: 8.8 percentage points, while in Argentina this difference is negligible (0.6%). So, the highest non-manual class for Spanish males has risen much more than for Argentine males. There is a reduction in male manual classes (V, VI, and VII) of 7.5 percentage points in Argentina, and 11.0 percentage points for Spanish males. These reductions are more relevant for skilled manual workers in Argentina and for unskilled manual in Spain. On the other hand, in Argentina the male petty-bourgeoisie (IV) decreases by 3%, while in Spain this drop is much larger: 8%.

The growth of the Spanish female non-manual classes (I, II, III) is also larger than in Argentina (40% versus 33%), and again Class I is responsible for the larger differences. The female petty-bourgeoisie (IV) decreased equally in both countries (around 14 percentage points), more so in rural areas, particularly in Spain. The manual classes (V, VII, VII) decreased more in Spain (25.5) than in Argentina (18.7), basically because of the higher decrease in the unskilled manual class in Spain. This is also more so for agricultural workers, and more markedly in Spain.

As a general tendency, the non-manual male and female classes have risen more in Spain than in Argentina, basically due to the growth of the highest service class (I). And the unskilled manual class has markedly decreased more in Spain than in Argentina. The male petty bourgeoisie declined less in Argentina than in Spain, with no differences in the case of women, with a decrease of around 14 points in each country.

In aggregate terms, this gives us a brief overview of the major changes that have taken place in the class structures of both countries for the 34 years covered by our study.

The rates of absolute intergenerational mobility for men and women in both countries are discussed in the next point. We rely on the usual stand of analyzing the change in social positions between “origins” (the father’s occupation during the respondent’s adolescence) and “destinations” (people in the sample’s occupation at the time of the survey).

Table 2. Changes in occupational structure between origins (parents’ classes) and destinations (sons/daughters’ classes) (a) Males and females, 25-59 years old, born between 1951 and 1985

Classes EGP I, II, III, IV, V+VI, VII	Males				Females
	Argentina		Spain		Argentina		Spain
	Origin	Desti-nation	Origin	Desti-nation	Origin	Desti-nation	Origin	Desti-nation
I: Higher managers and professionals	8.8	9.4	6.2	15.0	9.0	9.3	6.2	11.9
II: Lower managers and professionals	5.6	12.0	6.2	11.8	5.9	11.9	6.5	15.6
III: Intermediate occupations (non-manual)	8.7	12.1	9.3	13.9	9.4	35.8	10.0	35.1
IV: Small employers and own account workers	30.0	27.1	25.5	17.5	27.7	13.7	24.8	10.2
V+VI: Lower supervisory and technical occupations; semi-routine (manual) occupations	16.8	13.6	18.8	16.9	16.7	3.4	18.4	3.9
VII: Routine (manual) occupations	30.1	25.8	34.0	24.9	31.3	25.9	34.2	23.2
N	4171	4171	6948	6948	4084	4084	6740	6740
Total	100.0	100.0	100.0	100.0	100.0	100.0	100.0	100.0
(b) Both sexes, 25-59 years old, born between 1951 and 1985
Classes EGP I, II, III, IV, V+VI, VII			Both Sexes
			Argentina			Spain
			Origin	Destination		Origin	Destination
I: Higher managers and professionals	8.9	9.4	6.2	13.5
II: Lower managers and professionals	5.7	11.9	6.3	13.7
III: Intermediate occupations (non-manual)	9.0	23.8	9.7	24.3
IV: Small employers and own account workers	28.9	20.5	25.1	13.9
V+VI: Lower supervisory and technical occupations; semi-routine (manual) occupations	16.8	8.5	18.6	10.5
VII: Routine (manual) occupations	30.7	25.9	34.1	24.1
N	8255	8255	6948	6948
Total	100.0	100.0	100.0	100.0

Source: EU-SILC 2011 for Spain; CEDOP 2003-2012 for Argentina.

Absolute mobility

We briefly note, following Breen (2004, pp. 3-4) that a reference to absolute mobility, under the presence of a mobility table, is that “the cross-tabulation of class origins by current class position (or class destination) quite easily reveals patterns and rates of mobility, where mobility is understood simply as movement between origins and destinations.” In Table 3 we offer descriptive mobility results for six classes in Argentina and Spain, by sex.

Table 3. Absolute mobility results. Argentina and Spain, by sex

Absolute Mobility for six class scheme: I, II, III, IV, V+VI, VII	Argentina 25-59		Spain 26-59		Argentina 25-59	Spain 26-59
	Men	Women	Men	Women	Both Sexes	Both Sexes
Dissimilarity Index	10.5	32.7	19.0	40.0	21.5	29.3
Total observed mobility	67.5	74.7	69.1	75.9	71.0	72.5
Total non-vertical mobility (NVM)	18.9	23.4	15.6	21.9	21.1	18.8
Total vertical mobility (VM)	48.6	51.3	53.5	54.0	49.9	53.7
VM / NVM	2.6	2.2	3.4	2.5	2.4	2.9
Upward vertical mobility –UVM	28.9	30.4	35.9	37.2	29.7	36.5
Downward vertical mobility – DVM	19.6	20.8	17.6	16.8	20.2	17.2
UVM / DVM	1.5	1.5	2.0	2.2	1.5	2.1

Source: EU-SILC 2011 for Spain; CEDOP 2003-2012 for Argentina.

An important first observation is based on comparisons of the index of dissimilarity (ID) between total marginal in mobility tables. For men, this index (sometimes identified as “structural mobility”) shows for Spain a value that is almost twice that of Argentina, 22% in the case of women. That is, the change in “structural mobility” is more marked in Spain. And this structural mobility (dissimilarity index) for women is three times that of males in Argentina, and twice as much in Spain. This mobility is by far larger for women in both countries, due in part to the effects of occupational segregation as Salido (2001) noted for Spain.

The absolute mobility observed is near to 70% for males and around 75% for women in both countries. Vertical mobility⁷ is slightly higher in Spain, both for males (53.5% versus 48.6%) and females (54% versus 51.3%). Upward vertical mobility is 7 percentage points higher in Spain, for males and females. Hence, downward vertical mobility is slightly higher in Argentina. Consequently, the rate of upward over downward vertical mobility is higher in Spain (around 2.0 versus 1.5 for each sex).

Relative mobility

We now proceed with an exploration of relative mobility, that is, the chances of reaching one destination rather than another depending on origin. We begin our analysis by taking into account four birth cohorts -for both Argentina and Spain-, which constitute the main area of interest of this study. After that, and beyond analyses by cohorts, some considerations about international comparisons will be offered. Our efforts will basically rely on the analysis of strengths of relative mobility between origins and destinations.

We consider the birth cohort perspective for each country first, based on the analysis of relative mobility for men (first panel in Table 4a), and then for women (second panel in Table 4a). We also offer below Table 4b with results for both sexes.

Table 4. Relative mobility models by cohorts and sex. Argentina and Spain (a) Relative mobility by sex (1)

Models (6 Classes: I, II, III, IV, V+VI, VII)	ARGENTINA					SPAIN
	Men: N= 4171					Men: N= 6948
MEN	L2 (*)	d.f.	BIC	rL2 %	ID %	L2	BIC	rL2 %	ID %
Conditional independence [OC] [DC]	705.9	100	-127.69	----	15.3	1082.3	197.66	----	16.0
Constant association [OD] [OC] [DC]	72.5	75	-552.66	89.7	4.6	78.8	-584.62	92.7	3.9
Unidiff - Uniform Difference	71.6 p=0.82	72	-528.62	89.9	4.5	73.8 p=0.17	-563.14	93.2	3.7
Linear Unidiff	72.5 p=1.00	74	-544.34	89.7	4.6	75.4 p=0.06	-579.24	93.0	3.8
Regression Type Goodman -Hout	38.5 p=0.16	48	-361.62	94.5	3.1	43.6 p=0.13	-380.99	96.0	2.8
(2)
Cohorts	1951-1959	1960-1968	1969-1976	1977-1985	Linear	1951-1959	1960-1968	1969-1976	1977-1985	Linear
Unidiff Parameters	1.000	1.001	0.934	1.044	0.006	1.000	1.081	1.230	1.164	0.065
(3)
Models (6 Classes: I, II, III, IV, V+VI, VII)	ARGENTINA					SPAIN
	Women: N= 4171					Women: N= 6948
WOMEN	L2 (*)	d.f.	BIC	rL2 %	ID %	L2	BIC	rL2 %	ID %
Conditional independence [OC] [DC]	645.5	100	-186.32	----	12.6	744.8	-137.02	----	11.2
Constant association [OD] [OC] [DC]	82.8	75	-541.07	87.2	4.9	90.6	-570.74	87.8	3.9
Unidiff - Uniform Difference	81.5 p=0.72	72	-517.46	87.4	4.8	79.1 p=0.00	-555.78	89.4	3.7
Linear Unidiff	82.8 p=1.00	74	-532.76	87.2	4.9	79.2 p=0.00	-573.33	89.4	3.7
Regression Type Goodman -Hout	39.6 p=0.00	48	-359.72	93.9	3.4	49.3 p=0.03	-373.96	93.4	2.6
(4)
Cohorts	1951-1959	1960-1968	1969-1976	1977-1985	Linear	1951-1959	1960-1968	1969-1976	1977-1985	Linear
Unidiff Parameters	1.000	0.989	0.906	1.061	0.004	1.000	0.862	0.771	0.645	-0.116
(b) Relative mobility. Both sexes
Models (6 Classes: I, II, III, IV, V+VI, VII)	ARGENTINA					SPAIN
	N = 8,255					N = 13,688
TOTAL	L2 (*)	d.f.	BIC	rL2 %	ID %	L2	BIC	rL2 %	ID %
Conditional independence [OC] [DC]	1220.5	100	318.67	----	13.0	1651.0	698.62	----	13.0
Constant association [OD] [OC] [DC]	60.6	75	-615.74	95.0	3.0	84.4	-629.90	94.9	2.9
Unidiff - Uniform Difference	59.4 p=0.75	72	-589.90	95.1	2.9	81.2 p=0.36	-604.53	95.1	2.7
Linear Unidiff	60.6 p=1.00	74	-606.72	95.0	3.0	83.5 p=0.34	-621.34	94.9	2.8
Regression Type Goodman -Hout	31.0 p=0.33	48	-401.88	97.5	1.9	45.7 p=0.06	-411.45	97.2	2.0
Cohorts	1951-1959	1960-1968	1969-1976	1977-1985	Linear	1951-1959	1960-1968	1969-1976	1977-1985	Linear
Unidiff Parameters	1.000	0.968	0.934	1.025	-0.001	1.000	0.976	1.028	0.895	-0.024

(*) The p values presented with the deviances (L2) refer to the chi2 tests of differences

between distinct models and the constant association one.

Note: Given the presence of cells with 0 values for women, 0.1 was added to all cells for females in both countries.

Source: EU-SILC 2011 for Spain; CEDOP 2003-2012 for Argentina.

As usual, the baseline model of independence (or null model) does not produce a good fit. The constant association model produces a significant improvement in fit, with an 89.7% increase in explained association over the null model in Argentina and 92.7% in Spain. The constant association model produces a good fit in both countries for men. This model should be preferred with respect to the uniform difference (Unidiff) model in both nations.

We turn now to the results for women. As in the case of men, the constant association model produces a major advance, increasing the explained association by 92.7% in Argentina and by 87.8% in Spain. BIC notably increases its negative value, and the ID also decreases, more markedly in Argentina. No improvements were produced by the Unidiff model in Argentina, while in Spain it might be preferred, based on its BIC and ID values, plus the fact that the deviance test between both models (constant association and Unidiff) is significant in this country. The Unidiff parameters in Spain monotonically decline from the oldest to the youngest cohort. Unlike men, women seem to show high social fluidity for the origin-destination association over birth cohorts and, therefore, somewhat less dependence on their class origins with the passage of time. The negative value of the linear Unidiff parameter for Spain suggests a linearly decreasing O-D association.

Finally, for both sexes, the Regression Type model does not produce a good fit for men (and for both sexes) in spite of its low ID, given that its BIC value drastically decreases with respect to that of the previous models –as known, BIC attach more weight to parsimony-. The L2 test difference also suggests that this model does not improve on the previous ones in either country for men. However, for women the Regression Type model shows an improvement and it is statistically significant.

We briefly explore similarities and differences in relative mobility between Argentina and Spain -offering a direct comparison of mobility models between both nations (Table 5)- we note that the common association model for men and box sexes is the preferred model. Women are somehow closer to showing variability, and, in terms of Unidiff parameters, Argentina has lower social fluidity than Spain.

Table 5. Relative mobility models for comparison of Argentina and Spain

Models (6 Classes: I, II, III, IV, V+VI, VII)	SPAIN – ARGENTINA
MEN (N=11,119)	L2	d.f.	BIC	rL2 %	ID %
1. Conditional independence: OC, DC	1618.2	50	1152.39	----	15.2
2. Common association: OD, OC, DC	106.0	25	-126.93	93.4	3.7
3. Unidiff - Uniform Difference	105.7	24	-117.87	93.4	3.7
Countries	Spain	Argentina
Unidiff Parameters	1.000	0.971
Test of L2 (M2) – L2 (M3) for Men	p = 0.58
WOMEN (N=10,824)	L2	d.f.	BIC	rL2 %	ID %
1. Conditional independence: OC, DC	1250.1	50	785.60	----	11.0
2. Common association: OD, OC, DC	66.4	25	-165.82	87.2	2.5
3. Unidiff –Uniform Difference	62.6	24	-160.35	87.4	2.4
Countries	Spain	Argentina
Unidiff Parameters	1.000	1.140
Test of L2 (M2) – L2 (M3) for Women	p = 0.05
TOTAL (N=21,943)	L2	d.f.	BIC	rL2 %	ID %
1. Conditional independence: OC, DC	2746.8	50	2247.01	----	12.7
2. Constant association: OD, OC, DC	146.2	25	-103.71	94.7	2.9
3. Unidiff –Uniform Difference	144.8	24	-95.07	94.7	2.9
Countries	Spain	Argentina
Unidiff Parameters	1.000	1.053
Test of L2 (M2) – L2 (M3) for Both	p = 0.23

Source: EU-SILC 2011 for Spain; CEDOP 2003-2012 for Argentina.

Note: For models in this table, C denotes “Country”