Jump to ContentJump to Main Navigation
Social Mobility in Europe$

Richard Breen

Print publication date: 2004

Print ISBN-13: 9780199258451

Published to Oxford Scholarship Online: November 2004

DOI: 10.1093/0199258457.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2019. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 18 April 2019

Opportunities, Little Change: Class Mobility in Israeli Society, 1974–91

Opportunities, Little Change: Class Mobility in Israeli Society, 1974–91

Chapter:
(p.315) 13 Opportunities, Little Change: Class Mobility in Israeli Society, 1974–91
Source:
Social Mobility in Europe
Author(s):

Meir Yaish

Publisher:
Oxford University Press
DOI:10.1093/0199258457.003.0013

Abstract and Keywords

Engages with the long lasting debate in sociology concerning the consequences of industrialization process for social mobility. It is argued that Israeli society provides one of the most adequate tests of this thesis. In this context, then, the analysis in this chapter is guided by two main questions: (1) has equality of opportunity in Israeli mobility increased over time, (2) has the mobility process in Israeli society become more meritocratic over time. The analysis in this chapter is based on data from two nationally representative surveys that were tailored to the study of social mobility–the 1974 and the 1991 Israeli mobility surveys. It is shown that Israelis (men and women alike) experience high level of social mobility and fluidity–but with little temporal variations. It is also shown that while the Israeli stratification system has some meritocratic components, these did not gain in strength over time. Thus it is concluded that, in the Israeli context, social mobility, and the industrialization process do not go hand in hand.

Keywords:   equality of opportunity, industrialization, Israel, meritocracy, social fluidity, social mobility, temporal changes in social mobility

At the heart of the long lasting debate in sociology regarding the consequences of the industrialisation process for social mobility is the so-called ‘liberal thesis of industrialisation’. This states that the industrialisation process brings about not only more opportunities for social mobility, but also more equality of opportunity as social selection processes become more meritocratic (cf. Blau and Duncan 1967; Treiman 1970; Featherman and Hauser 1978; but see Sorokin 1959; Lipset and Zetterberg 1959; Featherman et al. 1975; Erikson and Goldthorpe 1992, for arguments against this thesis). In this context, Israeli society will serve as a case study to answer two main questions:

  1. (i) Has the Israeli mobility and fluidity pattern increased over time?

  2. (ii) Has the mobility process in Israeli society become more meritocratic over time?

It has previously been argued that in the context of the above-mentioned debate Israeli society, along with the Republic of Ireland (see Whelan and Layte in this volume), provides one of the most adequate tests of the industrialisation thesis (Yaish 1998, 2000). This is claimed on two grounds. First, Israeli society has undergone immense economic (as well as demographic) changes over a very short period of time. Second, high quality data are readily available to link these changes to social mobility. In what follows, then, I study changes over time (or lack of them) in Israeli mobility and fluidity patterns by comparing results from two nationally representative cross-sectional mobility surveys in Israel: the 1974 mobility survey (Matras and Weintraub 1977) and the 1991 mobility survey (Kraus and Toren 1992). Previous Israeli research on mobility has been dominated by the status attainment approach (see, among others, Semyonov and Kraus 1983; Smooha and (p.316) Kraus 1985; Kraus and Hodge 1990), and only a few studies have taken the tabular approach (Matras 1963; Zloczower 1973; Matras et al. 1975; Matras and Weintraub 1977; Tyree et al. 1979; Goldthorpe et al. 1997; Yaish 1995, 1998, 2000). A common element in the studies that have used the tabular approach is their exclusion of women from the analysis. These studies, moreover, display a wide range of variation in their analytical techniques, methodologies, populations studied, and ultimately the quality of their data. A classification of these studies along these dimensions would identify the following three ‘generations’ of social mobility research in Israel:
  1. 1. Studies from the 1960s and early 1970s. Mobility studies in this period used occupational categories between which only inflow and outflow percentages of Israeli society were examined. The majority of the studies in this period, however, were typically based on incomplete, and often dubious, data of only certain sections of Israeli society (cf. Matras 1963; Zloczower 1973; Matras et al. 1975; an exception in this period is the study by Matras and Weintraub (1977), who examined the mobility of all Israeli men).

  2. 2. Studies from the 1980s and 1990s. Compared with the previous generation, mobility studies in this period can be said to: (1) pay special attention to issues of data quality; (2) use methodologies that are tailored to the study of both relative and absolute mobility; (3) study the Israeli mobility process within a class structure; (4) examine Israeli mobility in a comparative perspective; and (5) examine changes in the Israeli mobility pattern over successive birth cohorts (Tyree et al. 1979; Yaish 1995; Goldthorpe et al. 1997).

  3. 3. Studies from the late 1990s (Yaish 1998, 2000). Mobility studies in this period improved on the previous generation by offering more systematic examinations of changes over time in Israeli mobility and fluidity patterns by comparing the results of a standardised analysis between two cross-sectional, nationally representative mobility surveys.

Despite the heterogeneity in techniques, methods, and data quality, a comparison between the results of these studies across ‘generations’ would suggest that the mobility of Israeli men is characterised by:
  1. (1) Constant absolute mobility rates over time;

  2. (2) An increase in upward mobility rates over time;

  3. (3) A fairly common fluidity pattern over time;

  4. (4) A constant fluidity level over time;

  5. (5) A very high level of fluidity in a comparative perspective.

This study is part of the third generation of mobility research in Israel as it offers a systematic examination of changes over time in Israeli class mobility and fluidity patterns. However, it differs from previous mobility studies of (p.317) Israeli society in three ways. First, the analyses are based on data that represent the entire Israeli population, men and women, at two distinct points in time. Second, the Israeli class mobility pattern is studied at both the individual and the household level. Finally, this study uses analytical techniques that examine the effect of individual characteristics (e.g. the individual's educational level) on the relationship between groups of individuals (e.g. the class fluidity pattern).

The Israeli context

The last hundred years or so have seen many changes in Israeli society. At the beginning of the Zionist enterprise in 1882, Israel–Palestine was, as it had been for centuries, poor and underdeveloped. The ‘industrialisation process’ began with the economic boom of the 1930s and the Second World War (Carmi and Rosenfeld 1974: 477). After Israel was established (in 1948) the process intensified and the structure of the Israeli economy changed dramatically (Kraus 1992). Equating industrialisation processes with economic growth (Wrigley 1972: 226), Fig. 13.1 shows that for the most part Israel has experienced sustained economic growth since it was established: GDP per capita has increased from 10,000 NIS in 1950 to about 50,000 NIS in 1999. Within this general trend, periods of extreme economic growth as well as recession periods can be identified (see also the unemployment figures in Fig. 13.2).

As a result of massive immigration in the early years of statehood, from 1951 through to 1953, the newly created state suffered from severe unemployment. In 1954, mainly due to reparation money received from Germany,

                   Opportunities, Little Change: Class Mobility in Israeli Society, 1974–91

Fig. 13.1. Real (1995) GDP per capita in Israel: 1950–99

(p.318) the economy entered a period of rapid economic growth that continued until 1965. By the end of this period the reparation payments had ended (Aharoni 1991), and in 1966 there was widespread unemployment and a significant drop in GDP (Aharoni 1991: 79). Economic growth began again only after the Six-Day War of 1967, and the period from 1967 to 1972 is characterised by exceptionally high and rapid economic growth (Remba 1971). From 1974 through to 1988 economic growth improved only marginally. Towards the end of 1988, another recession gripped the economy, coupled with a relatively high rate of unemployment. Economic growth resumed only in the early 1990s, but ended in the second half of the decade, due in part to the aftermath of the massive influx of immigrants from the former Soviet Union.

Israel's economic growth has been accompanied by marked structural changes in its various economic segments. The proportion of the population actively engaged in agriculture declined, industrial growth slowed down in the late 1970s, and services expanded substantially, especially in the public sector (Kraus 1992). From 1975 to 1984, one out of seven workers was employed in industry, with the remainder employed in services, half of them in the public sector (Aharoni 1991: 91). Israel's economic growth and, in particular, the expansion of the service industries (and the public sector) created a demand for labour which, in turn, increased employment opportunities for women. Studies have shown that Israeli women's participation rates in the paid economy are similar to those of women in other western societies (Ben-Porath and Gronau 1985; Haberfeld and Cohen 1998). Thus, for

                   Opportunities, Little Change: Class Mobility in Israeli Society, 1974–91

Fig. 13.2. Selected characteristics of the Israeli labour force

(p.319) example, the participation rate of Israeli women in the paid economy has increased over the years from about 25 percent in 1955 to about 50 percent in 2000 (see Fig. 13.2). As in most western societies, gender segregation in the Israeli labour force is high, and has hardly narrowed between 1972 and 1983 (Cohen et al. 1987). Thus, for example, most Israeli women work in white-collar occupations (Semyonov and Kraus 1983), and about half of all employed women work in the public sector (Kraus 1992; Haberfeld and Cohen 1998). And, as can be seen in Fig. 13.2, twice as many women work in part-time jobs as men (30 percent and 15 percent, respectively).

Israel has a highly centralised, state regulated economy. Ownership of economic resources is divided between the government, the Histadrut (the General Federation of Labour), and the private sector, with a growing tendency towards privatisation of government and Histadrut enterprises. Up to the early 1990s, the proportion of organised labour in Israel was amongst the highest in western societies—about 90 percent of the workforce. The Histadrut which is, inter alia, a labour union negotiates collective arrangements with the government and the private sector that cover most aspects of employment relations.

Industrialisation is not the only process to have affected the Israeli stratification structure: demographic changes have also played a major role. First of all, Israeli society is—to an exceptional degree—a society of immigrants. Successive waves of Jewish immigrants have entered the region since the Zionist movement was established in 1882, and the establishment of the state of Israel in 1948 encouraged growing numbers of Jews to enter the country. Indeed, from 1948 to 1990 some 2,031,800 immigrants (the majority of whom are Jewish) entered the country (CBS 1991: 43, Table 2.2).1

Second, and related to the above, Israeli society is multi-ethnic. It has been described (Lewin-Epstein and Semyonov 1986) as a dual society but on two different levels. At the primary level, there is the division between Jews and non-Jews, the latter being predominantly Arabs. Arabs holding Israeli citizenship make up around 17 percent of the national population. In regards both to area of residence and occupation, a high degree of segregation exists between citizen-Arabs and Jews. There is evidence (cf. Shavit 1992; Lewin-Epstein and Semyonov 1993) that Arabs have gained some benefit from geographical segregation, in that within their own enclaves they are less exposed to labour-market discrimination. At the secondary level, there is then the division among Jews between those of European or American origin (Ashkenazi Jews) and those of Asian or African origin (Sephardi Jews). The Sephardi are disadvantaged relative to the Ashkenazi in their opportunities (p.320) for socio-economic achievement, but there are indications that the degree of ethnic inequality in this respect is tending to weaken over time (cf. Kraus and Hodge 1990; Yaish 1998, 2001).2

Finally, it is important to recognise the role of the Zionist movement and government interventions in the development of Israel's economy. To begin with, the Zionist movement was an ideological movement that aimed to establish a new, modern, Jewish society in Palestine (Eisenstadt 1967: 2–3), and emphasised the ‘return to the soil’ of Jews in Palestine. By returning to the soil they meant to establish an agricultural basis for the Jewish community in Palestine (Eisenstadt 1967: 4). Indeed, within the first fifty years of the Zionist enterprise new agricultural settlements were established in Palestine. After Israel was established, the state took over the tasks of the Zionist movement. A policy of ‘decentralisation’ was adopted in which a large part of the population was settled on the periphery of the country, mainly in settlements with an agricultural infrastructure (Kipnis 1990). And, for primarily ideological reasons, new entrants to agriculture were offered tax relief and agricultural settlements were offered capital and other resources (Hanaki et al. 1989). At the same time, the establishment of the state of Israel marked a historical turning point for Palestinian society. Arabs became a subordinate, minority sub-population in Israel, as the new Jewish state confiscated some 40 percent of the land owned by them. This left most Israeli-Arabs landless, and thus contributed a great deal to the proletarianisation of Israeli-Arabs which began in the 1930s (cf. Rosenfeld 1978).

Data and variables

The analyses in this chapter are based on data from two nationally representative surveys that were tailored to the study of social mobility in Israel (see Table 13.1). The 1974 mobility survey (hereafter 1974MS) was conducted by the Israel Central Bureau of Statistics as part of its 1974 civilian labour force survey. The target population of the sample is all Israeli persons (Jews and Arabs) aged fourteen and over, excluding the institutionalised population. The sample includes 3,500 households, with 15,078 persons aged fourteen and above, of whom 11,917 are Jews and 3,161 Arabs (Matras and Weintraub 1977).3 The 1991 mobility survey (hereafter 1991MS) was conducted by Kraus and Toren (1992), and carried out by ‘Pori Research Institute’ in 1991. The target population of the sample is all Israeli persons (Jews and Arabs) aged eighteen and over, excluding the institutionalised population. The sample includes (p.321)

Table 13.1. Technical details of the data

Date

References for survey details

Survey organization

Ages

Mode

Population

Response rate (%)

House holds

1974

Matras and Weintraub (1977)

CBS

14+

Interview

Civilian L.F.

90

3,500

1991

Kraus and Toren 1992

Pori Agency

18+

Interview

All

86

5,800

5,800 households, with 9,926 persons aged eighteen and older, of whom 8,158 are Jews and 1,768 Arabs.4 The analyses in this chapter are based on two sub-samples of these data. Accordingly, from each mobility survey a sub-sample of all Israeli men and women aged 25–65 who were part of the Israeli civilian labour force is analysed.5

Three variables are at the heart of the following analysis—class origins, educational qualifications, and class destinations. However, differences exist between the surveys in the level of specificity of the information that is available for the construction of these variables. In what follows I explain the origins and implications of these differences.

The information on father's occupation in the 1974MS is classified into the two-digit Israeli occupational classification. The implication of this is that only the ‘standard’ seven-class version6 of the Goldthorpe class schema can be applied to this occupational classification (see Yaish 1998: ch. 3). To achieve comparability in the analysis of the two sub-samples I used the same occupational classification, and hence class classification, also in the 1991MS. Unlike father's occupation, the information on the respondent's current, or last, occupation is classified into the three-digit Israeli occupational classification. Thus, I was able to utilise the seven-class version that is appropriate for the study of women; that is, class IIIb is combined with class VIIa instead of class IIIa.7 The information on educational qualifications in the (p.322) 1974MS is limited to years of schooling only. The implication of this is that I am unable to apply the CASMIN educational variable to these data. Again, to achieve comparability in the analysis of the two sub-samples the effect of education on class mobility is assessed using years of schooling also in the 1991MS.

The intergenerational class mobility tables, for men and women in each survey, are then presented in appendix A to this chapter. In addition to these four mobility tables, I constructed two ‘complete mobility tables’ (also shown in appendix A). In complete mobility tables the household's, rather than the individual's, class position is the focus of the analysis. Accordingly, the class position of households of unmarried persons is determined by their own employment, while that of married couples takes into account the labour market experiences of both men and women in a household. In constructing this household class position I followed Erikson and Goldthorpe's (1992: 266) dominance approach.8

Trends in absolute mobility

Studying Israeli society enables us to follow trends in mobility rates in a society where most stages of the industrialisation process are covered by data of high quality. It is well established by now that the industrialisation process in Israel–Palestine began in the late 1930s, and that this process accelerated in the 1960s. The 1974MS and the 1991MS cover these periods well. The industrialisation process in Israel and other factors which I have reviewed earlier have, no doubt, altered the Israeli class structure.

Table 13.2 presents the marginal distributions in the Israeli mobility tables, by gender and year of survey. It shows how both between origins and destinations and over time (i.e. between the two surveys) the Israeli class structure has changed. Thus, for example, the dissimilarity indices, reported at the bottom of Table 13.2, between the origin and destination distributions in the 1974 mobility table are as high as 45 and 44 percent, for men and women respectively. In the 1991 mobility table, the respective indices reach 27 and 38 percent. For comparison, in the nations of the CASMIN project the dissimilarity index in the men's class distributions of only three nations (Hungary, Poland, and Japan) is higher than 27 percent, although all are lower than 45 percent (Erikson and Goldthorpe 1992: 193, Table 6.2). Among women, however, the Israeli dissimilarity indices are within the range of the (p.323)

Table 13.2. Marginal distribution of origin and destination class in the seven-class intergenerational mobility tables for Israeli men and women aged 25–64, and dissimilarity indices (Δ) derived from pairwise comparisons of origin, destination, and year of survey

Class

Origins

Destinations

1974

1991

1974

1991

Men

Women

Men

Women

Men

Women

Men

Women

I+II

6

11

11

15

15

26

23

32

III

6

8

7

8

6

9

8

14

IVab

53

58

31

31

20

16

18

6

IVc

15

6

11

5

5

4

2

1

V+VI

6

6

16

18

22

7

29

9

VIIa

11

11

21

22

31

38

19

37

VIIb

3

0

2

2

1

0

0

1

N

2,969

1,080

3,414

2,978

2,969

1,080

3,414

2,978

Men

Women

Comparison

1974

1991

1974–91

1974

1991

1974–91

Dissimilarity indices(Δ)

Origin/Destination

45

27

44

38

Origin/Origin

27

28

Destination/Destination

18

14

five CASMIN nations for which data on women's class distributions were available. The high indices of dissimilarity in Israel imply that a high rate of intergenerational class mobility is expected for both men and women. The question yet to be resolved is whether total rates of mobility have increased as Israel has become more industrialised.

The total mobility rates calculated from the Israeli mobility tables are at high, but nonetheless constant, levels. Thus, for example, among men the Total Mobility Rates (TMRs) are 74 and 73 percent, in 1974 and 1991 respectively. In a comparative perspective, with the CASMIN nations, Israel appears to be among the most mobile nations (along with Sweden and Hungary, 73 and 76 percent respectively; see Erikson and Goldthorpe 1992: 195, Table 6.3). Among women, the TMRs are 76 and 77 percent. Compared with the CASMIN nations, Israeli women do not appear to have as exceptionally high a level of total mobility as men (see Erikson and Goldthorpe 1992: 245, Table 7.4).9

(p.324)

                   Opportunities, Little Change: Class Mobility in Israeli Society, 1974–91

Fig. 13.3. Total mobility rates, total upward mobility rates, and total downward mobility rates in Israel, men and women aged 25–64 by birth-cohorts

A more detailed examination of these data supports this general pattern of no change in the Israeli TMR. Figure 13.3 presents graphs that represent ‘smoothed’ yearly measures (one year moving average) of the Israeli TMRs based on the consolidated datasets.10 The bold lines represent the TMRs of men (dark) and women (light). The dotted and broken lines then represent a decomposition of TMRs into total upward mobility and total downward mobility, respectively.

The trend that Fig. 13.3 reveals is very clear. The total rate of mobility has stayed fairly constant over time, for both men and women alike. One can say, then, that very little association exists between the industrialisation process and TMRs in Israel, as a comparison between Fig. 13.1 and 13.3 illustrates.

It is also apparent in Fig. 13.3 that throughout the period downward mobility is considerably more frequent among women than among men (see the broken lines). However, it would also appear that over time (for both men and women) downward mobility declines while upward mobility increases.11 For Israeli men, moreover, upward mobility is considerably more frequent than downward mobility. For women, on the other hand, upward mobility would appear to exceed downward mobility only half way through the period.

(p.325) What, then, are the theoretical implications of this analysis? To begin with, the analysis showed that for men and women alike the available opportunities for mobility within a changing class structure have a strong effect on class mobility. In this respect, the empirical evidence I have presented supports the expectation that in industrial nations more opportunities are available at the top of the class structure. Thus, for Israeli men and women alike upward mobility has increased over time while downward mobility has decreased. Second, upward mobility has become more frequent than downward mobility in Israeli society. However, Israeli women's mobility pattern, as in other industrial societies (cf. Erikson and Goldthorpe, 1992: ch. 7), has a strong (though declining) component of downward movement. The theoretical implication of this latter finding would appear to reject Braverman's (1974) argument that the participation of women in the labour force would somehow offset the long-term tendency for the working class in industrial society to decline.

Trends in relative mobility

Table 13.3 presents the results of fitting the ‘Constant Social Fluidity’ (CnSF) and the Unidiff models to the 1974 and the 1991 mobility tables, for men and women separately. The second row of Table 13.3 shows that Model B (CnSF) misclassifies fewer than 3 percent of all cases, for both men and women. The model, moreover, captures more than 93 and 86 percent of the association between father's class and respondent's class, for men and women respectively. What is more, in the case of Israeli men, the returned p-value is very close to the conventional standard of 0.05 (p = .04), while it is well above this standard for women (p = .38). Overall, then, it is possible to accept this model for both men and women. The implication of this analysis is that the Israeli

Table 13.3. Results of fitting the CnSF and the Unidiff models to the 1974 and the 1991 seven-class intergenerational mobility tables, Israeli civilian labour force aged 25–64, by gender

Model

d.f.

Men(N = 6,383)

Women(N = 4,058)

G 2

p

rG 2

Δ

G 2

p

rG 2

Δ

A. Ind. {OT}{DT}

72

761

.00

12.0

284

.00

8.7

B. CnSF {OT}{DT}{OD}

36

52

.04

93

2.4

38

.38

87

2.3

C. Unidiff

35

50

.05

93

2.5

36

.41

87

2.2

Notes: O= origin class; D= destination class; T= year of survey.

bic statistics (for Models A, B, and C, respectively) by gender: men: 130, −263, −257; women: −314, −261, −255.

(p.326) fluidity pattern (men and women alike) has not changed over time. It is important, next, to explore the possibility that within this apparent similarity, the strength of the OD association has changed over time.

To address this question, I apply the Unidiff model (see Xie 1992; Erikson and Goldthorpe 1992; Firth 1998) to these data. The third row of Table 13.3 shows that, for both men and women, the Unidiff test did not improve significantly upon the fit obtained by the previous model (G 2 is reduced by only 2 points for the loss of 1 degree of freedom). The implication is that the Israeli fluidity level (men and women alike) has not changed—let alone increased—over time.12

Since our examination of a trend in the Israeli fluidity pattern relies on two cross-sectional datasets from two points in time, it could be that the above analysis misses out some of what may have happened between these two points. Thus, a secular trend towards increasing fluidity may be better detected across successive birth-cohorts. To pursue such an examination entails, first, a consolidation of the two mobility surveys, and then to fit the Unidiff model to the mobility tables of as many birth-cohorts as possible (at most, there are fifty-eight yearly birth-cohorts). To obviate a potential problem of low cell counts in yearly birth-cohorts, however, I fitted the Unidiff model to the eighteen mobility tables that form the time series that was introduced earlier (see n. 10 above).

The patterns that emerge from the analysis of the above-mentioned time series are presented in Fig. 13.4 and 13.5, for men and women respectively. For Israeli men, then, Fig. 13.4 shows that the strength of the OD association has, with some fluctuation, remained relatively constant over time. For Israeli women, however, Fig. 13.5 shows that the strength of the OD association has tended to weaken over time, particularly since the 1916–55 mobility table.13

To summarise, then, in the context of the theoretical debate about the effect of the industrialisation process on social fluidity, the Israeli analysis would appear to lend support to the camp that rejects the liberal thesis. It was shown that the industrialisation process in Israel, which began in the 1930s, (p.327)

                   Opportunities, Little Change: Class Mobility in Israeli Society, 1974–91

Fig. 13.4. Unidiff parameter estimates: Israeli men

                   Opportunities, Little Change: Class Mobility in Israeli Society, 1974–91

Fig. 13.5. Unidiff parameter estimates: Israeli women

(p.328) does not appear to have affected the Israeli fluidity pattern. The analysis further throws doubt on the assumption that a causal (and functional) link exists between industrialisation and increasing fluidity levels. This is in spite of the finding that the fluidity level of Israeli women may have slightly increased over time. This is because, if we accept that industrialisation is the mechanism that causes greater fluidity in society, it should apply to men and women alike. However, the analysis showed that the fluidity level of Israeli men, and households, has not weakened over time.

Commonality and variation in social fluidity

The analysis above was concerned with very general (global) tests of propositions about changes over time in social fluidity. The objectives of the following analysis are to provide a more detailed description of the Israeli fluidity pattern, and to examine the ways in which this detailed pattern may have changed over time. These objectives can be reached by applying the so-called ‘core model’ of social fluidity (cf. Erikson and Goldthorpe 1992) to the Israeli mobility tables. Previous research in Israel (Goldthorpe et al. 1997; Yaish 1998, 2000) has shown that the core model captures, with few modifications, the fluidity pattern of Israeli men. The point of departure for the following analysis is thus the variant core models that were applied to and accepted for the Israeli data (see appendix B in which the Israeli variants of the core model are described).14 In what follows, I expand this analysis to Israeli women.

Applying the Israeli model: men

The results of fitting the Israeli variant models to the 1974 and the 1991 mobility tables of Israeli men are presented in panel A of Table 13.4.15 These models misclassify dissimilarity rider (D) in fewer than 3.3 percent of the cases, capture more than 90 percent of the association between origin and destination class, and the returned p-values from the models are higher than the conventional 0.05 level. On all accounts, then, these models fit the data well. It is possible now to examine the parameter estimates derived from these models, as presented in panel B, and to compare them to the core parameters estimated by Erikson and Goldthorpe (1992) using data from several European countries.

Several points of interest emerge from the comparison. Beginning with the parameters of the 1974 variant model (Model C), it is shown that all parameters (p.329)

Table 13.4. Results of fitting the core model of social fluidity to the seven-class intergenerational mobility table for Israeli men aged 25–64, by year of survey

Model

d.f.

1974 (N = 2,969)

1991 (N = 3,414)

G 2

p

rG 2

Δ

G 2

p

rG 2

Δ

A. Ind.

36

385

.00

11.4

376

.00

12.5

B. Core

28

95

.00

75

6.2

83

.00

78

6.0

C. Variant

28

37

.11

90

2.9

37

.12

90

3.3

D. C–n.s.a

30/29

44

.05

89

3.1

37

.15

90

3.3

Note: Bic statistics (for Models A, B, C, and D, respectively) by year of survey: 1974: 97, −129, −187, −196; 1991: 83, −145, −191, −199.

(a) C–n.s. = Model C omitting non-significant parameters.

Parameters estimated for the variant core model (s.e. in parentheses), by year of survey, and parameter values from core model given by Emkson and Goldthorpe ( 1992 :135)

Model

HI1

HI2

IN1

IN2

IN3

SE

AF1

AF2

1974

(C)

−0.23 (0.06)

−0.44 (0.12)

0.45 (0.10)

0.16 (0.13)

0.75 (0.37)

−0.48 (0.17)

−1.12 (0.59)

0.47 (0.06)

(D)

−0.24 (0.05)

−0.49 (0.11)

0.53 (0.07)

n.s.

0.79 (0.37)

−0.51 (0.17)

n.s.

0.48 (0.06)

1991

(C)

−0.16 (0.05)

−0.18 (0.08)

0.33 (0.08)

0.30 (0.11)

1.80 (0.75)

0.01 (0.37)

0.37 (0.12)

0.40 (0.05)

(D)

−0.16 (0.05)

−0.18 (0.08)

0.33 (0.08)

0.30 (0.11)

1.79 (0.26)

n.s.

0.37 (0.12)

0.40 (0.05)

Core

−0.22

−0.42

0.43

0.81

0.96

−1.03

−0.77

0.46

Note: Parameters in italics are modified as explained in appendix B.

achieve their expected sign. This suggests that the Israeli fluidity pattern, taken as a whole, follows the general expectation of the core model. Having said that, a deviation from the core level is observed with reference to three effects: (1) a non-significant IN2 effect; (2) a non-significant AF1 effect; and (3) a weak sectoral effect (SE). In what follows these deviations from the core level and their implications for the Israeli fluidity pattern will be discussed.

To begin with, the statistically non-significant IN2 effect in this variant model implies that the propensity for immobility among members of the Israeli service class, petty bourgeoisie, and farmer class is lower than the core level.16 Moreover, the parameter estimates (see Model Δ in panel B) indicate (p.330) that the propensity for immobility in these classes is about half the core level. For example: the additive odds representing the propensity for immobility of the service class and the petty bourgeoisie, where IN1 and IN2 apply, are 1.70 in Israel compared to a core level of 3.45 (e0.53 = 1.70 and e(0.43+0.81)=1.24 = 3.45 respectively). Likewise for the farmer class, where IN1, IN2, and IN3 apply, the additive odds for immobility are 3.74 in Israel compared to 7.24 in the core level (e1.32 = 3.74 and e1.98 = 7.24 respectively). This confirms the conclusions of other research that Israeli society is a highly ‘fluid’ society in a comparative perspective (Tyree et al. 1979; Goldthorpe et al. 1997; Yaish 2000). Second, the non-significant AF1 effect and the weak SE effect suggest that the boundaries between agricultural and non-agricultural classes are permeable in Israel. On the one hand, no specific barrier to movement between the farm-working class (VIIb) and the service class (I+II) appears to exist (the AF1 term is not significant). On the other hand, the general SE term in Israel reduces such mobility to only two thirds of what it would have been in the absence of this effect (e−0.51 = 0.60), compared with one third in the core model (e−1.03 = 0.36). The relatively easy intersectoral movement of labour in Israel may reflect the Zionist ideology and government intervention as outlined earlier. Interestingly, a similar pattern is found in Hungary (Erikson and Goldthorpe 1992: 153) where government intervention has played an important role in enforcing such a pattern.

Moving next to the parameter estimates from the 1991 variant model, Table 13.4 shows that all parameters in the variant model (Model C) achieve their expected sign, as in 1974. However, some deviations from the core pattern and the 1974 pattern are apparent. First, the relatively weak SE term in 1974 is no longer significant in 1991. However, the now statistically significant AF1 term (representing the barrier to movement between the farm-working class (VIIb) and the service class (I+II)) suggests that a specific barrier for intersectoral mobility exists in the 1991 mobility table. Thus, some intersectoral barriers for mobility have always existed in Israel—although these barriers have always been very weak. As mentioned earlier, the relatively easy intersectoral movement in Israel may be the result of Zionist ideology and government intervention (cf. Yaish 2000).17 Second, the pattern of immobility has changed between 1974 and 1991. This can be seen with reference to the IN2 effect—the effect that enhances immobility for the service class (p.331) and petty bourgeoisie. The IN2 term has a significant effect in the 1991 model, while it has no significant effect in the 1974 model. This piece of evidence suggests that the Israeli fluidity pattern has shifted closer to the core level over time. Finally, comparing the inheritance effects between this model and the core model reveals two distinctive features. On the one hand, the propensity for immobility of the service class and petty bourgeoisie, where IN1 and IN2 apply, is only about half of the core level (e0.63 = 1.88 and e1.24 = 3.45 respectively). On the other hand, inheritance among Israeli farmers is stronger than the core level by a factor of one-and-a-half (e2.42 = 11.2 and e1.98 = 7.24 respectively). Here again we see the distinctive position of the agricultural sector—this time its elite—in the Israeli class structure. Finally, hierarchical barriers to mobility in Israel seem to be slightly lower than the core level. It is calculated that short-range mobility, where HI1 applies, reaches nearly nine-tenths of what it would have been in the absence of such an effect (e−0.16 = 0.85), and long-range mobility, where HI1 and HI2 apply, reaches about two thirds of what it would have been in the absence of such effects (e−0.34 = 0.71).18 The equivalent figures under the core model are four-fifths (e−0.22 = 0.80) and just over a half (e−0.64 = 0.52), respectively.

Applying the Israeli model: women

The results of fitting the variant models to the 1974 and the 1991 mobility tables of Israeli women are presented in panel A of Table 13.5. The 1974 model misclassifies (Δ) little more than = percent of all cases, captures 67 percent of the association between origin and destination class, and the returned p-value from the model is higher than the conventional .05 level. Although the model performed better in the case of men, the overall performance of the model in the case of women led me to accept it in their case as well.19 A similar conclusion is reached regarding the 1991 variant model. Although, as can be seen in Table 13.5, the returned p-value from the model does not reach the conventional 0.05 level (p = .03), the model misclassifies fewer than = percent of cases, and captures 75 percent of the association between origin and destination class. Accepting the same models for men and women suggests that men and women in Israeli society share the same fluidity pattern.20

(p.332)

Table 13.5. Results of fitting the core model of social fluidity to the seven-class intergenerational mobility table for Israeli women aged 25–64, by year of survey

Model

d.f.

1974 (N = 1,080)

1991 (N = 2,978)

G 2

p

rG 2

Δ

G 2

P

rG 2

Δ

A. Ind.

36

104

.00

9.7

179

.00

8.4

B. Core

28

41

.05

61

6.1

55

.00

69

4.6

C. Variant

28

34

.21

68

5.2

44

.03

75

3.5

D. C–n.s.

n.a./30

n.a.

46

.03

74

3.7

Note: Bic statistics (for Models A, B, C, and D, respectively) by year of survey: 1974: −147, −155, −162, n.a.; 1991: −109, −169, −180, −194.

* C–n.s. = Model C omitting non-significant parameters.

Parameters estimated for the variant core model (s.e. in parentheses), by year of survey, and parameter values from core model given by Erikson and Goldthorpe ( 1992 :135)

Model

HI1

HI2

IN1

IN2

IN3

SE

AF1

AF2

1974

(C)

−0.09 (0.12)

−0.10 (0.16)

0.20 (0.20)

0.44 (0.22)

0.18 (1.10)

−0.73 (0.51)

−0.08 (1.06)

0.29 (0.10)

1991

(C)

−0.12 (0.07)

−0.26 (0.08)

0.19 (0.09)

0.16 (0.12)

−0.26 (0.69)

−0.67 (0.21)

0.42 (0.11)

0.13 (0.06)

(D)

−0.14 (0.06)

−0.26 (0.08)

0.25 (0.08)

n.s.

n.s.

−0.64 (0.17)

0.44 (0.11)

0.15 (0.06)

Core

−0.22

−0.42

0.43

0.81

0.96

−1.03

−0.77

0.46

Note: Parameters in italics are modified as explained in appendix B.

It is possible now to examine the parameter estimates derived from these variant models. Beginning with the 1974 survey, Model C in Table 13.5 shows that all the parameters achieve their expected sign. This suggests that the Israeli women's fluidity pattern, taken as a whole, follows the general expectation of the core model. However, only two parameters in the 1974 model are statistically significant (IN2 and AF2), and it may well be that with a larger sample size the statistical inference from this analysis fares better.21 But even if one is prepared to accept these parameters, with a very low level of certainty, it is still apparent that their size is considerably lower than the core (p.333) and the men's parameters (see Table 13.4). This, then, points to the fact that Israeli women enjoy a relatively high level of fluidity compared with men (see also note 21 above).

Moving next to the 1991 variant model (Model C). Most of the parameter estimates from this model, reported in panel B of Table 13.5, achieve their expected sign. Now, however, the majority of the parameters are statistically significant, and their size would appear to have shifted closer to the core level.22 This trend is also evident for Israeli men. However, several interesting differences do exist between women's and men's fluidity patterns. First, the relatively strong inheritance effect amongst male farmers (IN3) does not appear to apply for women. In fact, Israeli women of farm origins do not appear to benefit from intergenerational transmission of land and farms (IN3 is very weak and not statistically significant). This is not a surprising finding, as in Jewish agrarian communities (Moshavim) it is possible for families to divide their farm only to the benefit of a continuing son who will then become a farmer in that community. In the historically more agrarian Arab community an additional factor may be at work. It is well documented that most Arab farmers had lost their land to the newly established Jewish state, resulting in a very low level of class inheritance amongst farmers (cf. Yaish 1998, 2001).

Second, the general inheritance of class position (IN1)—and particularly inheritance of farm, bourgeois, and service class positions (IN2)—would appear to be weaker among women than men. This points to the fact that the level of women's fluidity is generally higher than men's. As women's class position is partly determined by marriage (cf. Erikson and Goldthorpe 1992), it is expected that it will not be strongly related to their father's class position. Finally, the SE term for mobility is statistically significant for women but not for men—although it is still lower than the core level.

The results of the above analyses can be easily summarised as follows:

  1. (1) The Israeli fluidity pattern is well captured by the core model in the 1974 and 1991 mobility tables;

  2. (2) Israeli men and women share a basically similar fluidity pattern—as it is captured by similar core models; however,

  3. (3) Israeli women's level of fluidity is higher than that of men—as it is captured by the size of the core parameters;

  4. (4) Overall, fluidity in Israel appears to be at a higher level compared to the core level; however,

  5. (5) The Israeli fluidity pattern has shifted closer to the core pattern over time.

(p.334) In sum, I showed that social institutions and other particular features of Israeli society have a limited though distinctive influence on the Israeli fluidity pattern (see, further, Yaish 1998, 2000). Nonetheless, I was also able to show that the Israeli fluidity pattern would appear to oscillate around (and even shift closer to) a common fluidity pattern that characterises industrial nations. It is within this cross-national commonality that the Israeli fluidity level would appear to be near the most fluid end.23

Ascription and achievement in the process of social fluidity

Educational qualifications are often viewed as the most important determinant of occupational success (cf. Shavit and Müller 1998). It is also argued that the educational system is a sorting mechanism in society (see e.g. Collins 1979), which replaces the more traditional selection mechanisms such as the extended family and the juridical system (see e.g. Sorokin 1964: 169–70). It is clear from the above that the effect of education on social mobility is related to the way the educational system of society is organised. As the subject of this study is Israeli society, I shall begin with a review of Israel's educational system.

Israel's educational system is highly centralised and standardised: the Ministry of Education certifies teachers, develops curricula, sets the matriculation exams, and maintains the administration. And most schools and all universities are publicly funded.24 The Israeli education system has undergone two main stages of development since the founding of the state of Israel. Soon after independence, a compulsory education law was introduced (the ‘Compulsory Education Law’ 1949); then, in 1968, the entire Israeli school system was reformed.25 The objectives of the educational reform were not (p.335) merely ‘cosmetic’—to change the structure of the educational system. The reformers aimed to change the process of educational attainment in Israel by adopting an equal opportunity approach that would give each pupil the chance to fully develop his/her potential.

In sum, then, no a priori structural inequalities would appear to exist within the Israeli school system. This is because the Israeli educational system is standardised and centralised by the Israeli Ministry of Education. Second, reform of the Israeli education system in the late 1960s aimed to equalise educational opportunities between individuals from different social backgrounds. These features represent a clear government intervention that aims to bring Israel closer to the meritocratic model. To this, proponents of the liberal thesis would add that the industrialisation process itself should also promote a meritocratic society because in industrial society achievement replaces ascription in the process of class reproduction. Within this framework, moreover, merit is equated with education.

Having said that, the following analysis aims to estimate the extent to which, if at all, the association between class origins and destinations is mediated through education. The point of departure in this analysis is that the association between origins and destinations is fairly adequately estimated by the eight parameters of the variant core model that was fitted to the Israeli mobility tables earlier in this chapter. Accordingly, I introduce to the variant core model measures of individual educational attainment, and examine the effects of so doing on the parameters initially estimated.26 To the extent that these parameters shift towards zero, the association between class origins and destinations can be regarded as being mediated by the educational attainment variable that was introduced. Before we proceed, however, I should remind the reader that due to data limitations, as explained in the introduction, the educational variable in the multinominal logit models is years of schooling.27 Having made these preliminary clarifications, we can embark on the analysis.

Table 13.6, then, presents the results of fitting the Israeli variant core model to men's mobility tables in 1974 and 1991. Model C is identical to the log-linear model (Model D) in Table 13.4, and Model Δ adds the education variable to this variant model. It can be seen that the inclusion of the education variable, in both 1974 and in 1991, is statistically significant (for 6 d.f. G 2 is reduced by more than 1,000 points). This implies that education mediates some of the association between class origins and destinations. Put in other words, education has an important and significant role in the allocation (p.336)

Table 13.6. Results of fitting the core model of social fluidity to the seven-class intergenerational mobility table (controlling for education) for Israeli men aged 25–64, by year of survey

Model

1974 (N = 2,957)

1991 (N = 3,367)

d.f.

G 2

Δd.f.

δG 2

d.f.

G 2

δd.f.

ΔG 2

A. null

17,736

9,861.7

20,196

10,966.0

B. var

17,728

9,514.4

8

347.3

−0,188

10,645.0

8

321.0

C. B − n.s

17,730

9,521.0

2

6.6

−0,189

10,645.0

1

0.0

D. B + ed

17,724

8,390.7

6

1,130.3

−0,183

9,371.4

6

1,273.6

E. D–n.s

17,727

8,394.7

3

4.0

−0,185

9,372.0

2

0.4

Note: Bic statistics (for Models A, B, C, D, and E, respectively) by year of survey: 1974: −131,883, −132,167, −132,176, −133,258, −133,278; 1991: −153,061, −153,317, −153,326, −ndash;154,550, −154,566.

* D–n.s. 5 Model D omitting non-significant parameters.

Parameters estimated for the accepted model (s.e. in parentheses), by year of survey, and parameter values from core model given by Erikson and Goldthorpe ( 1992 :135)

Model

HI1

HI2

IN1

IN2

IN3

SE

AF1

AF2

1974

(C)

−0.24 (0.06)

−0.49 (0.11)

0.53 (0.07)

n.s

0.79 (0.37)

−0.51 (0.17)

n.s.

0.48 (0.06)

(D)

−0.07 (0.06)

0.02 (0.14)

0.48 (0.07)

n.s.

0.71 (0.36)

−0.29 (0.17)

n.s.

0.26 (0.07)

(E)

n.s.

n.s.

0.53 (0.06)

n.s.

1.10 (0.19)

n.s.

n.s.

0.24 (0.06)

1991

(C)

−0.16 (0.05)

−0.18 (0.08)

0.33 (0.08)

0.28 (0.11)

1.73 (0.26)

n.s.

0.39 (0.12)

0.40 (0.05)

(D)

−0.02 (0.06)

0.26 (0.10)

0.39 (0.08)

0.07 (0.11)

1.77 (0.26)

n.s.

0.30 (0.14)

0.30 (0.06)

(E)

n.s.

0.26 (0.10)

0.43 (0.05)

n.s.

1.78 (0.24)

n.s.

0.30 (0.14)

0.30 (0.05)

Core

−0.22

−0.42

0.43

0.81

0.96

−1.03

−0.77

0.46

Note: Parameters set in italics are modified as explained in appendix B.

of Israeli men to their respective class position. However, the parameter estimates from this Model (D), presented in the second panel, indicates that not all the association between class origins and destinations—as this is captured by the eight core parameters—is mediated by education. Furthermore, education does not appear to have a uniform effect on the core parameters. Whereas, for example, inheritance effects (IN1 and IN3) do not appear to move closer to zero once education is controlled for, hierarchical barriers to mobility (HI1 and HI2) would appear to be eliminated almost completely. (p.337) This suggests that education is important for a certain kind of mobility, while less so for another.

An examination of changes over time in the effect of education on the Israeli fluidity pattern, by comparing the core parameters from Models E between 1974 and 1991, reveals a remarkably stable pattern. To begin with, the three statistically significant parameters in the 1974 model (IN1, IN3, and AF2) remain statistically significant in the 1991 model. What is more, the size of each of these parameters has not changed, statistically significantly, over time.28 The implication of this is that the effect of education on the OD association, as far as these three core parameters are concerned, is fairly constant over time. At the same time, it can be seen that in 1991 two parameters that were not statistically significant in 1974 are statistically significant (HI2 and AF1). It appears, furthermore, that in the 1991 model the HI2 term has become positive once education is controlled for. That is, the effect of education on social fluidity has changed over time such that it not only offsets long-range hierarchical barriers for mobility (as Model E in 1974 indicates), but also advances such mobility (as Model E in 1991 indicates). However, the statistically significant AF1 term in the 1991 model would appear to offset part of this temporal change. All of the above would then suggest that, although education mediates a large part of the OD association in Israeli society, its effect on this association has remained fairly constant over time. Put in other words, the stratification process in Israeli society, as far as the fluidity pattern of Israeli men is concerned, has not become increasingly meritocratic over time.

The same analysis is repeated for women (see Table 13.7). As in the men's analysis, the inclusion of the education variable, in both 1974 and in 1991, is statistically significant (for 6 d.f. G 2 is reduced by more than 500 points). However, the effect of education on social fluidity is different for men and women, particularly in 1974. Thus, for example, the inclusion of the education variable in the 1974 mobility table would appear to mediate completely the core parameters. In particular, the only two significant terms in Models B (IN2 and AF2) are no longer statistically significant. What is more, the hierarchical barriers to mobility, although not significant in either of models B or C (i.e. before or after the inclusion of the education variable), have become positive. The implication of this is that education not only promotes hierarchical mobility for women, but it also reduces the effect of class inheritance.29

In the 1991 mobility table, the effect of education on the core parameters is similar to the men's case. To begin with, education has an important and (p.338)

Table 13.7. Results of fitting the core model of social fluidity to the seven-class intergenerational mobility table (controlling for education) for Israeli women aged 25–64, by year of survey

Model

1974 (N = 1,072)

1991 (N = 2,922)

d.f.

G 2

Δd.f.

ΔG 2

d.f.

G 2

Δd.f.

ΔG 2

A. null

6,426

3,323.8

17,526

8,621.6

B. var

6,418

3,254.0

8

69.8

17,518

8,492.7

8

128.9

C. B − n.s

n.a.

17,520

8,494.1

2

1.4

D. B + ed

6,412

2,725.9

6

528.1

17,514

7,192.6

6

1,302.0

E. D–n.s

n.a.

17,517

7,194.4

3

1.8

Note: Bic statistics (for Models A, B, C, D, and E, respectively) by year of survey: 1974: −41,533, −41,526, n.a., −42,012, n.a.; 1991: −131,236, −131,301, −131,316, −132,570, −132,592.

Δd.f. 5 Change in degrees of freedom; ΔG 2 = Change in G 2.

Parameters estimated for the accepted model (s.e. in parentheses), by year of survey, parameter values from core model given by Erikson and Goldthorpe ( 1992 :135)

Model

HI1

HI2

IN1

IN2

IN3

SE

AF1

AF2

1974

(B)

−0.07 (0.13)

−0.10 (0.16)

0.23 (0.20)

0.44 (0.22)

0.18 (0.55)

−0.70 (0.55)

−0.07 (1.09)

0.29 (0.10)

(D)

0.10 (0.13)

0.15 (0.19)

0.24 (0.21)

0.01 (0.25)

0.40 (1.15)

−0.54 (0.55)

1.30 (1.14)

0.08 (0.12)

1991

(C)

−0.14 (0.07)

−0.27 (0.08)

0.24 (0.08)

n.s.

n.s.

−0.59 (0.19)

0.44 (0.11)

0.12 (0.06)

(D)

−0.03 (0.07)

0.11 (0.10)

0.13 (0.09)

n.s.

n.s.

−0.59 (0.19)

0.25 (0.14)

0.00 (0.06)

(E)

n.s.

n.s.

0.13 (0.06)

n.s.

n.s.

20.59 (0.19)

0.27 (0.13)

n.s.

Core

−0.22

−0.42

0.43

0.81

0.96

−1.03

−0.77

0.46

Note: Parameters in italics are modified as explained in appendix B.

significant role in the allocation of Israeli women to their respective class position. However, not all of the association between class origins and destinations—as this is captured by the eight core parameters—is mediated by education. Furthermore, education does not appear to have a uniform effect on the core parameters. Whereas, for example, hierarchical barriers to mobility (HI1 and HI2) are no longer significant once education is controlled for, the inheritance effect (IN1) is—although it appears to move closer to zero. This suggests, as in the case of men, that education is important for a certain (p.339) kind of mobility, while less so for another. It also suggests that, over time, the effect of education in mediating class origin effects on Israeli women's fluidity pattern might have weakened. In particular, in 1991 education seems no longer to mediate the effect of class inheritance completely, as it did in 1974. Thus, one may argue, the stratification process in Israeli society, as far as the fluidity pattern of Israeli women is concerned, might have become less meritocratic over time.

As I have indicated earlier, however, the relatively small N in the 1974 women's mobility table would make an over time comparison of their core parameters somewhat problematic. In light of this, I am inclined to take a rather conservative approach in interpreting these results, and argue that the stratification process in Israeli society, as far as the fluidity pattern of Israeli women is concerned, has not become more meritocratic over time.

Conclusions

This chapter engaged with the debate concerning the consequences of the industrialisation process for social mobility and fluidity. Two contradictory views form this debate. On the one side, it is argued that a strong link exists between industrialisation and social mobility (cf. Treiman 1970). On the other side, it is argued that a link between industrialisation and social mobility does not exist (cf. Lipset and Zetterberg 1959; Sorokin 1959; Featherman, et al. 1975; Erikson and Goldthorpe 1992, among others). Studying mobility trends in the context of a unique society such as Israel sheds new light on this debate. Three issues were addressed in the analysis: (1) the extent to which absolute mobility has changed over time, (2) the extent to which relative mobility has changed over time, and (3) the extent to which education mediates relative mobility.

The analysis showed that for men and women alike the available opportunities for mobility within a changing class structure have a strong effect on class mobility. In this respect, the empirical evidence I have presented supports the expectation of the liberal thesis that in industrial nations more opportunities are available at the top of the class structure. Thus, it was shown that, for Israeli men and women alike, upward mobility has increased over time while downward mobility has decreased over the same period. At the same time, gender differences in mobility patterns were observed. In particular, the Israeli women's mobility pattern, as in other industrial societies (cf. Erikson and Goldthorpe 1992: ch. 7), has a strong (though declining) component of downward movement. This, in turn, would appear to reject a Marxist claim (see Braverman 1974) that the participation of women in the labour force would somehow offset the long-term tendency for the working (p.340) class of industrial society to decline. Finally, the analysis showed that the Israeli total rates of mobility (men and women alike) have stayed nearly constant over the years covered by these data. This constancy runs contrary to the liberal thesis explanation. Thus, it is concluded that social mobility and the industrialisation process do not go hand in hand (cf. Sorokin 1959).

The examination of changes over time in relative mobility in Israel led to a similar conclusion. That is, in the context of the theoretical debate about the effect of the industrialisation process on social fluidity, the Israeli analysis damages the liberal thesis that a causal (and functional) link exists between industrialisation and increasing fluidity. In spite of a nearly monotonic increase in Israel's GDP, the fluidity level of Israeli men has stayed nearly constant over time, while that of women has only recently increased slightly.

Nonetheless, the Israeli fluidity pattern (for men and women alike) has changed over time. I have shown that particular historical and political features of Israeli society may have affected the Israeli fluidity pattern. These changes, moreover, are mostly associated with land distribution (cf. Yaish 1998, 2000). That is, changes over time in the Israeli fluidity pattern do not correspond to the industrialisation process. At the same time, and in spite of the above, I have shown that there appears to be some force that pulls the Israeli fluidity pattern (men, women, and ultimately households) closer to a common level of fluidity. This common level and pattern of fluidity, moreover, is well captured by the core model of social fluidity developed by Erikson and Goldthorpe (1992).

Finally, the analysis examined the effect of education on class mobility in Israel. It was shown in this analysis that education has a very specific, and often limited, effect on the mobility process. In the Israeli case, it would appear, education affects mainly hierarchical mobility. Intergenerational inheritance of class position, for example, is not affected by education. This, then, would suggest that the Israeli stratification system may have some meritocratic components. However, it is also apparent that non-meritocratic components of the Israeli stratification system are common. More importantly, the non-meritocratic elements do not appear to have weakened over time.

The implications of this study go beyond the immediate concerns of Israeli sociologists and members of Israeli society. The study has value, I believe, for all students of social mobility and stratification in industrial nations. The results of this chapter serve to reject the idea that economic progress affects the process of stratification in industrial nations. In that respect, exponents of the liberal thesis offer a very functional and simple explanation, with little empirical support, for what should happen in the industrialisation process. Accordingly, stratification processes affect individuals' actions, but are beyond the control of individuals; their only determinant is the industrialisation (p.341) process and its logic. Once society internalises this logic, which is imposed on society, the process of stratification is affected.

A rejection of this scenario implies that other factors affect the stratification process. These factors, moreover, would appear to be common to all industrial societies and fairly constant over time. In light of the above, then, it is argued here that the stratification process is chiefly affected by individuals' actions—to fulfil their aspirations, to maintain their social standing, and/or in response to external factors. This being the case, then, the driving force behind social mobility is social—that is, individual-action.

Appendix A

Table 13.A1. Counts and marginals in the seven-class intergenerational mobility table Israelis aged 25–64, in the 1974 IMS (top) and 1991 IMS (bottom) surveys

Class of origin

Class of destination

I + II

IIIa

IVab

IVc

V + VI

IIIb + VIIa

VIIb

N

%

Men's mobility tables

I + II

80

11

19

7

31

25

173

6

178

33

46

9

69

45

380

11

IIIab

48

27

26

1

38

43

1

184

6

83

21

29

1

64

31

1

230

7

IVab

218

89

382

51

340

481

8

1,569

53

217

83

273

16

283

199

1

1,072

31

IVc

40

20

78

79

67

161

4

449

15

83

16

74

44

73

97

1

388

11

V + VI

34

13

29

12

62

35

1

186

6

96

38

79

7

227

92

1

540

16

VIIa

35

14

44

5

98

133

4

333

11

126

77

104

7

244

164

2

724

21

VIIa

3

2

16

6

14

30

4

75

3

10

13

12

1

25

19

80

2

N

458

176

594

161

650

908

22

793

281

617

85

985

647

6

%

15

6

20

5

22

31

1

23

8

18

2

29

19

Women's mobility tables

I + II

57

9

15

1

8

24

1

115

11

225

56

25

2

28

98

434

15

IIIab

33

11

4

3

5

34

90

8

89

35

11

2

18

83

1

239

8

IVab

125

53

120

23

46

254

2

623

58

273

132

59

9

92

354

8

927

31

IVc

12

3

9

12

5

19

60

6

61

20

9

4

11

48

6

159

5

V + VI

17

5

11

3

6

27

69

6

136

87

35

66

198

3

525

18

VIIa

32

11

16

1

4

54

118

11

155

90

28

6

61

301

7

648

22

VIIb

1

3

4

5

6

4

1

4

25

1

46

2

N

276

92

175

44

74

415

3

944

426

171

24

280

1,107

26

%

26

9

16

4

7

38

32

14

6

1

9

37

1

Complete mobility tables

I + II

146

10

22

6

32

28

244

8

428

37

39

9

67

50

0

630

17

IIIab

40

53

22

3

38

46

1

203

7

74

115

28

2

50

55

1

325

9

IVab

207

98

405

51

316

491

5

1,573

51

215

95

275

20

248

259

3

1,115

30

IVc

34

17

78

82

62

164

4

441

14

66

16

66

40

60

95

3

346

9

V + VI

32

13

29

12

74

38

1

199

6

96

40

72

4

200

119

1

532

14

VIIa

30

13

44

3

95

183

3

371

12

115

72

96

6

189

250

2

730

19

VIIb

2

2

16

6

14

31

5

76

2

7

12

13

2

21

18

0

73

2

N

491

206

616

163

631

981

19

1,001

387

589

83

835

846

10

%

16

7

20

5

20

32

1

27

10

16

2

22

23

0

(p.342)

(p.343)

Table 13.A2. Design matrices for the Israeli variants of the core fluidity model

1. Matrices identical to original core model and common to 1974 and 1991

                   Opportunities, Little Change: Class Mobility in Israeli Society, 1974–91

2. Affinity matrices for 1974

                   Opportunities, Little Change: Class Mobility in Israeli Society, 1974–91

3. Affinity matrices for 1991

                   Opportunities, Little Change: Class Mobility in Israeli Society, 1974–91

Appendix B

The Israeli model of social fluidity

Three alterations to the original core model were suggested for the 1974 mobility table (Goldthorpe et al. 1997; Yaish 1998, 2000; see Table 13.A2), all of which concern the positive affinity effect (AF2)—the effect that reinforces the linkage (p.344) between particular classes in the mobility table. It was then suggested that:

  1. (1) the pair of cells indicating mobility between the petty bourgeoisie, class IVab, and the service class, I + II, should be omitted from the AF2 term;

  2. (2) the cell indicating mobility from the petty bourgeoisie, class IVab, to the unskilled working class, VIIa, should be included in the AF2 term;

  3. (3) the cell indicating mobility from the skilled manual class, V + VI, to the unskilled working class, VIIa, should be omitted from the AF2 term.

The modified model then suggests that the Israeli petty bourgeoisie have a distinctively low propensity for mobility into the service class, and a distinctively high propensity for mobility into the unskilled working class. This was previously explained by the different composition of the Israeli petty bourgeoisie (class IVa vis-à-vis class IVb) compared to the CASMIN nations, coupled with the disadvantaged economic position of the Israeli self-employed. This model also suggests that the expected high propensity for mobility from the skilled working class (V + VI) into the unskilled working class (VIIa) does not exist in Israel. As with the petty bourgeoisie, the differences in the relative size of classes V and VI compared to the CASMIN nations, coupled with hierarchical differences between the two sub-classes, explain the Israeli variant fluidity pattern of class V + VI as a whole when compared to the core pattern.

Also for the 1991 mobility table three alterations to the core model were suggested (Yaish 1998, 2000; see Table 13.A2). These modifications follow a similar pattern to the 1974 variant model, with two differences. On the one hand, unlike the 1974 variant model, no evidence for a distinctively high movement of men from the petty bourgeoisie into the unskilled working class is found. It was argued that as a result of compositional changes (i.e. an increase in the share of class IVa within class IVab) the fluidity pattern of the petty bourgeoisie (i.e. class IVab) has shifted so as to be closer to the fluidity pattern of small employers (IVa). Thus, men from petty bourgeois origins have a lower propensity for mobility into the unskilled working class (VIIa) in 1991 than in 1974. However, even in 1991, the composition of the Israeli petty bourgeoisie (class IVab) remains quite distinct from that of the CASMIN nations, and they still do not have a positive affinity with the service class, as the core model expects. On the other hand, a new barrier for mobility from the skilled working class into the service class was suggested. This barrier, it was then proposed, is due to the exclusion of army personnel from the analysis. All in all, then, it was suggested that:

  1. (1) the pair of cells indicating mobility between the petty bourgeoisie, class IVab, and the service class, I + II, should be omitted from the AF2 term;

  2. (2) the cell indicating mobility from the skilled manual class, V + VI, to the unskilled working class, VIIa, is omitted from the AF2 term;

  3. (3) the cell indicating upward mobility from the skilled workers (V + VI) to the service class (I + II) is included in the AF1 term.

Notes:

(1) In this respect, Tyree et al. (1979), argue that immigration and social mobility are positively correlated. One can argue, thus, that the nearly constant influx of Jewish immigrants into Israel may have affected social mobility to such an extent that industrialisation is less relevant to the understanding of mobility trends in Israeli society. Against this I would argue that immigration and economic growth (i.e. industrialisation) are also very closely related; massive numbers of new arrivals into a society can initiate economic growth. By contrast, emigration may hinder economic growth (see e.g. Goldthorpe's (1992) analysis of Irish society). It is beyond the scope of this chapter to address the ‘immigration’ issue in Israel. Studies that have taken the ‘immigration’ issue to the heart of the analysis show, however, that, among Jewish Israeli men, immigrants and natives share a very similar fluidity pattern (Goldthorpe et al. 1997; Yaish 1998, 2000, 2002).

(2) There is very little reason to believe that one ‘mobility regime’ adequately characterises all sections of Israeli society. Nonetheless, previous studies show that men of the three main ethnic groups in Israeli society share a very similar fluidity pattern (Goldthorpe et al. 1997; Yaish 1998, 2001).

(3) For sampling procedures, see CBS, Labour Mobility Survey 1977, No. 544.

(4) The sampling procedures of this survey are similar to those of the CBS labour mobility surveys.

(5) I restrict the sample to respondents aged twenty five or over (rather than age twenty as is nearly the standard in most mobility research) to allow them the necessary time after compulsory military service to have completed their post-secondary studies and enter the labour market. I also omitted from the analysis employees in the non-civilian labour force. This is because, as mentioned above, the 1974MS includes information on the civilian labour force only. Thus, to allow comparability between the two surveys I restrict also the 1991MS population to the civilian labour force only.

(6) By the standard version I mean that it is not possible to separate class IIIa from class IIIb, although this separation is desirable when women are included in the analysis (see Erikson and Goldthorpe 1992).

(7) It is important to note that: (1) women dominate class IIIb, and (2) class origin refers to father's class position—also in the case of women. Therefore, in the context of intergenerational class mobility (particularly relative mobility), applying a less detailed classification to class origins should not have severe consequences. Having said that, it is possible to assess the severity of this problem based on the detailed information available from the 1991MS. Such an analysis indicated that the effects of different classifications of class origins on the mobility pattern of either men or women in the 1991MS are minimal.

(8) ‘Work time’ is taken as the first dominance criterion, so an employed person dominates one who is unemployed, while, among the employed, full-timers dominate part-timers. When partners still remain undifferentiated dominance is determined by the ‘work position’ criterion, so higher-level employment dominates lower-level employment (here I used the dominance 1 approach; see Erikson and Goldthorpe 1992: 266).

(9) In the complete mobility tables the TMRs are 69% and 65%, in 1974 and 1991, respectively.

(10) From the consolidated data I constructed a time series consisting of eighteen, forty-year birth-cohorts. That is, for each gender I constructed eighteen mobility tables comprising the following birth-cohorts: 1910–49, 1911–50, 1912–51, 1913–52, 1914–53, 1915–54, 1916–55, 1917–56, 1918–57, 1919–58, 1920–59, 1921–60, 1922–61, 1923–62, 1924–63, 1925–64, 1926–65, 1927–66.

(11) The same pattern emerges from the complete mobility tables. Thus, for example, between 1974 and 1991 upward mobility has increased from 22 percent to 30 percent, while downward mobility has decreased from 22 percent to 17 percent.

(12) The conclusion that the pattern of the OD association is constant over time was confirmed also when the complete mobility tables were analysed. Thus, for example, the p-value from the CnSF model is .122 (G 2 of 46 with 36 d.f.), and the Unidiff test did not improve significantly upon the fit obtained by this model (G 2 is reduced by about 2 points for the loss of 1 d.f.).

(13) Since the time series is constructed in such a way that many of the observations are included in more than one mobility table, I repeated the cohort analysis for three independent successive birth-cohorts (Israeli men and women who were born between 1909 and 1926 in the 1974MS; Israeli men and women who were born between 1927 and 1949 in both mobility surveys; and Israeli men and women who were born between 1950 and 1966 in the 1991MS). The results of this analysis confirmed my earlier conclusion. Whereas, for example, the CnSF model appears to fit the data well for both men and women (the model misclassifies fewer than 5 percent of the cases, and the returned p-values are well above the .05 level), the Unidiff model improves significantly upon the fit obtained by this model only for women (G 2 is reduced by 6 points for the loss of 2 d.f, p < .05). Under this model it is estimated that amongst women the strength of the OD association in the 1927–49 cohort is 61% of that in the 1910–26 cohort, while a further decline in the strength of this association is apparent in the 1950–66 cohort.

(14) As explained elsewhere (Yaish 1998, 2000), there are two variant core models in the Israeli case: a variant model for 1974 mobility table and a variant model for the 1991 mobility table.

(15) The results reported in Table 13.4 are slightly different from those presented by Yaish (1998, 2000). This is due to the fact that the destination class categories were modified in the current analysis to allow the inclusion of women.

(16) Since the majority of Israelis from petty bourgeois origins are sons of immigrants, it may be expected that they will have a relatively high propensity for mobility (e.g. Tyree et al. 1979 provide a discussion on the effect of immigration on social fluidity; see also Yaish 1998: ch. 7, 2002). This being the case, the non-significant IN2 term in the Israeli variant model may refer to the petty bourgeoisie only. However, a model that excludes the IN2 term from the cell indicating immobility of the petty bourgeoisie did not confirm this hypothesis, and the IN2 term for immobility of the service and the farmer classes was still not significant.

(17) Goldthorpe et al. (1997) report a similar finding. These authors, however, give two alternative explanations for this feature. First, this deviation is explained by a pro-agricultural ideology in Israel which aims to create a Hebrew farmer. Second, this deviation may be the result of a small sample size which, in particular, affects the number of individuals in the agricultural sector (see Goldthorpe et al. 1997: 9).

(18) Note, however, that in the 1991 model the AF1 term is statistically significant. Since the AF1 term refers to a combination of hierarchical and sectoral barriers to mobility, it may offset the hierarchy (particularly HI2) and the SE parameters.

(19) It is worth noting that the core model was originally developed to characterise the class fluidity pattern of men. In fact, the model was developed based on the mobility tables of French and English men (cf. Erikson and Goldthorpe 1992: ch. 4).

(20) Indeed, applying the CnSF and the Unidiff models to men and women's mobility tables, for each survey separately, confirmed this suggestion. This analysis also reveals that the main difference between the two genders, as far as fluidity is concerned, is that the OD associations in the women's mobility tables were estimated (under the Unidiff model) to be weaker, by a factor of 0.7, than in the men's tables. This would then imply that the ‘core’ parameter estimates from the women's tables should be smaller than those estimated from the men's tables.

(21) Note the very small sample size (N = 1,080). The very small sample size also resulted in unstable estimates. Thus when I started removing non-significant parameters from the model, the size—and standard errors—of the remaining parameters changed substantially.

(22) The very small sample size in the women's 1974 mobility table makes a more detailed examination of changes over time in women's fluidity patterns problematic.

(23) It should be further noted here that the Israeli variant core models achieved an acceptable fit also to the 1974 and the 1991 complete mobility tables. An examination of the parameter estimates of these models showed how closely these resemble men's parameters, particularly in 1974. This is not surprising, as women's participation in full-time employment in the Israeli labour force is significantly lower than men's (see Fig. 13.2). This being the case, the majority of Israeli households' class position is determined by the husband's class position. Indeed, only about 15% in 1974 and 30% in 1991 of Israeli households' class positions are determined by women's class position. Details of these analyses are available from the author on request.

(24) At the same time, Israel's educational system is differentiated along several lines. First, there are separate Jewish and Arabic schools. Second, within the Jewish system, there are several streams catering to different shades of religious orthodoxy (Kraus et al. 1998).

(25) Prior to this reform, attendance at school was compulsory by law until age fourteen (or completion of the eighth grade), and the Israeli school system was a 1 + 8 + 4 system. As part of the reform the minimum leaving age was raised to fifteen (with tuition free until age sixteen), and the Israeli school system was changed to a 1 + 6 + 3 + 3 system. As of today, however, the Israeli school system consists of an almost equal mixture of the pre- and post-reform structures (see Kraus et al. 1998 and the citations therein).

(26) This procedure simply requires that I rewrite the topological log-linear model (i.e. the Israeli variant core model) as a multinomial logit model for individual-level data (cf. Breen 1994; Logan 1983).

(27) Given my data constraints, years of schooling should be viewed as a proxy measure of educational qualifications for the purpose of estimating how much the latter mediates the direct OD association.

(28) This conclusion is reached after a t-test was used to test each pair of these parameters. The results of this analysis are available from the author on request.

(29) Note, however, the very small N in this mobility table. For this reason the results of the analysis should be read with some caution.