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, 2018. 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: 21 January 2019

Changes in Intergenerational Class Mobility in Hungary, 1973–2000

Changes in Intergenerational Class Mobility in Hungary, 1973–2000

Chapter:
(p.287) 12 Changes in Intergenerational Class Mobility in Hungary, 1973–2000
Source:
Social Mobility in Europe
Author(s):

Péter Róbert

Erzsébet Bukodi

Publisher:
Oxford University Press
DOI:10.1093/0199258457.003.0012

Abstract and Keywords

Investigates temporal changes in Hungarian mobility patterns. Large-scale data sets of the Hungarian Central Statistical Office, collected between 1973 and 2000 are used for this purpose. In addition to descriptive statistics, log-linear and log-multiplicative models are fitted to the data in order to investigate trends of temporal changes. Descriptive results indicate that the restructuring of the class distribution slowed down in the 1980s in comparison to the 1970s but it increased again in the 1990s. Observed mobility rates turned out to be relatively high but data does not indicate an increase in the openness of the Hungarian society. For relative mobility rates, the hypothesis of constant social fluidity cannot be rejected for Hungary. Though an increase in social fluidity did occur between 1973 and 1983, it levelled off between 1983 and 1992, and it reversed between 1992 and 2000.

Keywords:   absolute and relative mobility rates, class inheritance, cohort analysis, Hungary, log-linear models, social fluidity, social mobility, social structure

Hungary has an exceptionally rich set of social mobility data. A question on father's occupation was included in the 1930 Census and again in the 1949 Census. Later, the Hungarian Central Statistical Office (CSO) carried out large-scale social mobility surveys in 1962–4, 1973, 1983, and 1992. These data sources have recently been complemented by the Way of Life and Time Use Survey, carried out by the CSO in 2000, where a question on father's occupational class was also included. This chapter uses data from the last four of these surveys and focuses on trends in class mobility over nearly three decades in Hungary.

Since the available Hungarian mobility data covers the whole twentieth century, with its historical, economic, and political changes, Hungary always served as a good test case for the long standing debate in sociology about the role of industrialisation and democratisation in increasing mobility chances. In this respect, the effects of industrialisation and of political measures on long-term changes in status attainment seem to reinforce each other, as a recent analysis confirmed (Luijkx et al. 2002). But we also tend to agree with Ossowski (1963) who proposes that the proper question to ask is not whether Communism increased mobility but whether the decline of class boundaries was any greater than might have been expected from economic developments. In this respect, the hypothesis formulated by Kelley and Klein (1977, 1981) about the gradual reclosure of the social structure in socialist societies at a certain point after the Communist takeover sounds plausible. Based on earlier findings about the Hungarian mobility regime, which we summarise below, we also expect that the secular trend of increasing openness was more characteristic of the beginning of the period we consider here; and this phase was followed by a levelling off (if not a decline) in mobility chances.

(p.288) In the next sections we first summarise the findings of previous studies on social mobility in Hungary as well as the structural and institutional developments of the last three decades. Second, we provide a description of changes in absolute mobility rates between 1973 and 2000. Third, we present the results of statistical models of relative mobility rates fitting the constant social fluidity model (CnSF) and the uniform difference (Unidiff) models to the data. This part of the chapter is supplemented with a quasi cohort analysis in order to get a better view of changes over time. We also provide a Hungarian version of the core fluidity model. Finally, we summarise the findings and draw our conclusions.

Social mobility in Hungary in the light of previous studies

Our first analyses are based on the 1973 Hungarian social mobility survey. Andorka (1983), using this same dataset, presented mostly descriptive statistics on absolute mobility rates as well as the occupational profiles of successive birth cohorts based on job history data. He argued that mobility connected to self-employment and managerial status had been influenced mainly by socialist political measures. Simkus (1981) found that the increase of intergenerational mobility in Hungary was largely based on the decline of intergenerational inheritance of farming as well as of higher status non-manual occupations. Decomposing structural transformation revealed that the increase in social fluidity was mainly due to forced mobility (Simkus 1984).

Focusing on long-term tendencies of change in mobility chances, Andorka (1990a, b) investigated data within a longer time-span between 1930 and 1983. On the level of observed mobility rates Hungarian society became more open between the 1940s and the 1960s. This trend did not persist into the 1970s. For almost the same historical period, Ganzeboom et al. (1991) performed a careful test of trends in men's social fluidity between 1930 and 1989. In addition to CSO data, data from TARKI (Social Research Informatics Centre) for the years 1982, 1986, 1988, and 1989 were included in order to extend the time-span. Occupational classifications in the various data-files were handled more rigorously, different levels of aggregation were applied, and various types of log-linear model were fitted to the data. The authors concluded that the Hungarian mobility regime gradually opened up over time and that this development continued even after 1973. As a slight exception to this general trend, the increase of inheritance of self-employment at the early stage of Communism could be mentioned.

Wong and Hauser (1992), however, came to another conclusion after analysing the 1983 Hungarian social mobility data. In this analysis, they (p.289) investigated intergenerational fluidity based on father's occupation and son's and daughter's first occupation, computing twenty 8 × 8 mobility tables, one for each sex in ten five-year birth cohorts covering about fifty years. Although the increase in social fluidity was substantial for the older cohorts who entered the labour force in the early years of the socialist transition, the authors found support for the relative closure thesis for the younger cohorts. As for gender differences, occupational inheritance was smaller for women than for men but women's mobility chances were more strongly determined by their social origin than were men's.

Some of the most recent trend analyses of social mobility in Hungary have included the 1992 social mobility data. Andorka (1997) compared social fluidity between 1973 and 1983 as well as between 1983 and 1992. For the first time-span he found significant changes in mobility chances for men but this was not the case for the second. For women, no changes in mobility chances could be observed for any of the two periods. Most recently, Szonja Szelényi (1998: ch. 4) attempted to give a summary of Hungary's mobility processes during the twentieth century, based on CSO data from 1983. Her log-linear analysis applying row and column effects models with freely scaled distances, fitted to the data of four Hungarian birth cohorts, suggested ‘no evidence of massive changes’ despite the ‘radical interventionist policies of socialism’ (Szelényi 1998: 70). Temporal changes seemed to be closer to the trendless fluctuation proposed by Sorokin (1959). Only a minor ‘socialism effect’ was found, indicating increased exchange mobility between the managerial and working classes, but even this process produced more deterioration for the managers than improvement for the workers. The results provided no support for the claim that offspring of small proprietors were discriminated against by the socialist regime, their inheritance parameters being roughly constant over time.

Since access to Hungarian social mobility data became available relatively early, it has been widely used for international comparisons (Hazelrigg and Garnier 1976; Tyree et al. 1979; McClendon 1980a, b; Heath 1981; Grusky and Hauser 1984). These papers were based on three by three father to son mobility tables and the limitations of the data as well as serious concerns about their comparability have been expressed by Hazelrigg and Garnier (1976: 500) and—more sharply—by Erikson and Goldthorpe (1992: 27).

The most elaborate analysis by Grusky and Hauser (1984), testing the Featherman Jones Hauser (FJH) hypothesis (Featherman et al. 1975), applied log-linear and log-multiplicative models to the three by three father to son mobility tables, adding also exogenous variables to the models for measuring industrialisation, educational enrolment, social democracy, and income inequality. In a comparative view, Hungary turned out to be an outlier: the final models—in order to make them fit—had to include a specific dummy variable for Hungary. As a proof of the presence of a ‘socialism effect’, in a subsequent paper by Hauser and Grusky, where Hungary together with (p.290) Czechoslovakia, Poland, and Yugoslavia represented the ‘Eastern block’ among twenty-two nations, the authors found that ‘exchanges between manual and non-manual sectors take place 27 percent more frequently in socialist countries than in their non-socialist counterparts’ (Hauser and Grusky 1988: 738).

Comparing Hungary to a less industrialised western country, where modernisation and economic transformation took place in the same historical period, Simkus et al. (1990) reached the conclusion that deviations in mobility patterns between Ireland and Hungary were largely a consequence of changes in the private ownership of land and capital in the latter country. In fact, Andorka and Zagorski (1980) came to the same conclusion when comparing Hungary and Poland, based on data from the early 1970s. They found that differences in mobility patterns were mostly due to the fact that agriculture became collectivised in Hungary, while private farming survived in Poland under Communism.

Erikson and Goldthorpe (1992) also analysed data from the 1970s in The Constant Flux and found the Hungarian mobility pattern to be strongly affected by the nationalisation of agriculture which led to increased absolute mobility rates. Hungary deviated from the common pattern in many respects, such as a higher rate of recruitment to all classes from farm origins; a lower rate of recruitment from skilled worker-class origins; a lower rate of outflows from class origins to service-class positions; and a higher rate of outflows to skilled or unskilled working class positions. With respect to relative mobility rates, Erikson and Goldthorpe developed a core social fluidity model based on the concept of hierarchy, inheritance, sector, and affinity effects. When testing the significant differences between the effect parameters estimated for the core model and the nine nations separately, Hungary produced four differences (out of the eight effects): the HI1 effect was stronger, while the IN1, SE, and AF2 effects were weaker. As this review of the literature indicates, results from previous research on the Hungarian mobility regime are not univocal. For international comparative analyses, Hungarian ‘exceptionalism’ seems to be a typical feature; Hungary seems to deviate both from the western nations and from other former state-socialist countries. Some conclusions of the earlier studies focusing on temporal changes in social mobility in Hungary emphasised a trend toward increasing openness. Other studies, however, reported a more marked reversal in the increase of mobility opportunities. This makes our analysis highly relevant as we have a very recent dataset from 2000.

Economic and institutional changes in Hungary during the last three decades

In the period between the early 1970s and 1990s, the basic feature of economic and political developments was the slowdown of the ‘revolutionary’ (p.291) pace of political changes (which had started in the late 1960s). The quota system in education was abolished in the beginning of the 1960s and the general political climate of the country started to become more liberal after 1963, the year of political amnesty for the last 1956 prisoners. At the same time, the pace of industrialisation began to decline and the weaknesses of the planned economic system became more apparent. A moderate liberalisation of the economy and the introduction of certain market elements were expected to improve economic conditions and the New Economic Mechanism (NEM) was introduced in 1968 to reach these goals. In fact, previous studies on long-term changes in mobility processes and status attainment have proposed 1968 as the most important historical time-point when a new economic and political era started in Hungary (Luijkx et al. 1998, 2002).

Despite attempts at reforming the planned system, the serious weakness of the Hungarian economy became obvious in the 1980s. Economic development continued to deteriorate; the annual growth of GDP varied between only 1 and 3 percent during most of the decade. The Communist party had to make unpopular decisions, such as increasing prices, which led to growing inflation rates. After one and a half decades of political liberalism and political legitimacy based on the continuous improvement of living conditions, the political opposition intensified its activity, building on the increased dissatisfaction of the Hungarian population due to declining material circumstances. The process culminated in founding new political parties and the Communist regime had to start negotiations with the opposition (the Round Table Discussions). This led to the establishment of the parliamentary multiparty system and to free elections in 1990, just at the time when socialism collapsed in all other central and eastern European countries.

The 1990s brought significant economic changes to Hungary. First of all, the collapse of the system resulted in a decline in economic performance. The main features of the economic disruption were the absolute decrease in GDP (in 1991 there was a 12 percent decline), the two-digit inflation rate, and the high level of unemployment (12 percent in 1993). Since the economic crisis affected the mining, manufacturing, and steel industry sectors most strongly, the labour market situation deteriorated faster for men than for women. The growing service sector (trade, transport, communication, and services) provided more job opportunities for women and thus men became over-represented among the unemployed. Due to privatisation, the other main feature of economic developments after 1990, the proportion of the self-employed increased from 3 to 10 percent of the labour force and the majority of employees started to work for the private sector. At the same time the employment rate dropped, becoming lower than in most of the developed industrial societies at around 42 percent for men (aged 15–59 years) and 30 percent for women (aged 15–54 years). A slow economic improvement started only in the second half of the 1990s. By the end of the decade, both the unemployment (p.292) and inflation rates turned out to be less than 10 percent, and an increase of 3–4 percent per annum in GDP was reached (Fazekas 2000).

In sum, the period our data cover can be characterised by a continuously declining performance of the economy. The 1990s also brought a decline in the tendency of modernisation as the collapse of socialism resulted in deterioration in the institutions of the safety net and in the support for those coming from disadvantaged families. This latter argument also supports our expectation about a levelling off or a reversal in the trend of increasing social fluidity in Hungary.

Trends in origin and destination class structures

The economic and institutional shifts in Hungarian society over the last three decades are reflected in the class distributions for the four points in time, presented in Table 12.1 for men and women. (Information on data and variables can be found in appendix A to this chapter). The crucial transformation, which has occurred to the class structure of Hungarian males and females between the early 1970s and 2000, indicates an overall trend towards upgrading of work in Hungary. Growing managerialism and professionalism and the emergence of the service sector have led to three major structural changes. First, the proportion of individuals belonging to the service class (and primarily to the upper service class) increased, especially between 1973 and 1983; second, the percentage of unskilled and agricultural workers decreased; and, third, the petty bourgeoisie increased in line with the new market conditions that emerged in the 1980s and accelerated in the 1990s.

Dissimilarity indices provide a condensed view of shifts over time. The figures in Table 12.2 also show that there were more structural changes between 1973 and 1983 than in the subsequent part of the period studied, especially between 1983 and 1992. However, economic restructuring and the emerging private sector in the 1990s also resulted in a stronger shift in the class structure as the higher Δ for 1992/2000 reveals.

The gender-specific differences in the class distributions reveal a marked segregation in the Hungarian class structure. Although the total magnitude of the service class is larger for women, men have a higher share in the upper service class. As in most modern societies, the routine non-manual class is strongly feminised in Hungary. Men, on the other hand, dominate the skilled manual worker class.

The pattern of dissimilarity indices is also gendered. The structural changes in class destinations between 1973 and 1983 were much stronger for women but they seem to be less affected by the system transformation in the 1990s. For men the origin–destination Δs in Table 12.2 show a decreasing trend between 1973 and 2000; that is, there are less and less structural changes (p.293)

Table 12.1. Origin and destination class structures in 1973, 1983, 1992, and 2000 (males and females aged 20–69 currently in employment or unemployed having had a job)

1973

1983

1992

2000

Origin

Destination

Origin

Destination

Origin

Destination

Origin

Destination

Malesa

Class I

2.5

5.5

4.2

10.6

5.3

10.0

9.6

11.2

Class II

3.1

8.5

5.5

9.3

7.2

10.2

7.1

11.4

Class IIIa

2.8

3.9

2.2

2.5

1.8

1.8

2.0

2.5

Class IVab

6.9

1.7

4.8

2.5

4.5

6.2

4.4

9.6

Class IVc

26.4

0.7

15.8

1.2

7.4

1.5

2.1

2.6

Class V + VI

16.6

34.1

19.2

36.6

24.8

37.2

32.8

32.2

Class VIIa + IIIb

20.7

32.1

27.5

28.9

28.3

26.5

26.2

26.0

Class VIIb

21.0

13.6

20.9

8.5

20.8

6.6

15.8

4.5

Total

100.0

100.0

100.0

100.0

100.0

100.0

100.0

100.0

Femalesb

Class I

3.1

2.3

4.6

6.0

5.8

5.6

9.8

8.1

Class II

3.2

9.9

5.8

20.3

8.5

23.4

7.7

26.4

Class IIIa

3.1

22.9

2.6

16.5

1.7

19.6

1.7

17.1

Class IVab

8.1

1.0

4.1

1.4

4.4

3.6

4.1

6.2

Class IVc

24.5

0.4

15.5

0.2

6.7

0.4

2.0

0.7

Class V + VI

18.7

8.9

19.8

14.5

25.1

11.7

30.9

11.9

Class VIIa + IIIb

21.6

37.6

27.1

32.2

27.4

31.6

27.4

28.2

Class VIIb

17.7

17.0

20.4

8.9

20.3

4.2

16.4

1.4

Total

100.0

100.0

100.0

100.0

100.0

100.0

100.0

100.0

(a) Values of N for males in the year: 1973 = 11,221; 1983 = 9,047; 1992 = 7,212; and 2000 = 2,609.

(b) Values of N for females in the year: 1973 = 8,271; 1983 = 7,814; 1992 = 6,516; and 2000 = 2,134.

(p.294)

Table 12.2. Dissimilarity indices (Δ)

1973/1983

1983/1992

1992/2000

1973/2000

Destination/destination

Males

9.2

5.6

7.6

18.0

Females

20.1

8.4

8.6

30.8

Origin/origin

Males

13.4

9.2

12.5

32.8

Females

13.7

9.8

9.2

29.2

1973

1983

1992

2000

Origin/destination

Males

37.8

29.3

21.8

12.1

Females

42.5

34.9

36.9

37.0

Males' destination/females' destination

29.3

28.7

36.0

31.8

hidden behind the observed mobility patterns. For women, the structural modifications dropped between 1973 and 1983 but they began to increase slightly between 1983 and 2000. Gender segregation in the class structure of men and women is also displayed by the larger origin–destination Δs for women (where fathers and daughters are compared). The dissimilarity indices for the destination classes of men and women (Table 12.2) do not indicate any decline either. Of course, Δs are rough measures but it seems that gender specific structural differences in the class distribution are persistent in Hungary despite large shifts in the female class structure between 1973 and 2000.

Trends in absolute mobility rates

We start to investigate the historical trend in intergenerational class mobility in the context of absolute rates. First of all, we consider the total mobility rate (TMR), which is the percentage of respondents found in the off-diagonal cells of the mobility table. In other words, this measure represents the percentage of men and women whose destination class was different to their class of origin. Furthermore, we decompose the TMR into total vertical (TV) and total non-vertical (TNV) mobility rates. Vertical mobility refers to all such cases where moves occur between lower-level and higher-level classes; while non-vertical mobility means movements between classes within the same level. Finally, total vertical mobility is decomposed into total upward (TU) and total downward (TD) mobility. These rates are defined in accordance with Erikson (p.295)

Table 12.3. Absolute class mobility rates in 1973, 1983, 1992, and 2000 (seven-class schema) (males and females aged 20–69 currently in employment or unemployed having had a job)

1973

1983

1992

2000

Males

Total mobility rate

75.1

72.4

69.7

63.8

Total vertical

45.1

52.0

53.8

50.3

Total non-vertical

30.0

20.4

15.9

13.5

Total vertical/total non-vertical

1.5

2.5

3.4

3.7

Total upward

35.1

40.7

39.9

33.2

Total downward

10.0

11.3

13.9

17.1

Total upward/total downward

3.5

3.6

2.9

1.9

Mobility into the service class (I + II)

11.9

16.7

15.8

15.0

Mobility into the self-employed class (IVab)

1.3

2.2

6.0

9.0

Mobility into the unskilled worker class (VIIa + IIIb)

30.9

27.3

24.1

23.3

N

11,221

9,047

7,212

2,610

Females

Total mobility rate

80.2

77.7

75.3

73.9

Total vertical

42.2

51.1

45.6

57.3

Total non-vertical

38.0

26.6

29.7

16.6

Total vertical/total non-vertical

1.1

1.9

1.5

3.4

Total upward

26.6

39.4

30.8

40.1

Total downward

15.6

11.7

14.8

17.2

Total upward/total downward

1.7

3.4

2.1

2.3

Mobile into the service class (I + II)

10.2

22.5

23.8

28.8

Mobile into the self-employed class (IVab)

0.8

1.2

3.4

6.0

Mobile into the unskilled worker class (VIIa + IIIb)

37.2

30.7

29.3

25.6

N

8,271

7,814

6,516

2,134

and Goldthorpe (1992: ch. 6). These measures are displayed in Table 12.3, for males and females in employment and aged 20–69.

Observed total mobility rates tend to decrease in Hungary, more for men than for women. The decline is quite gradual for women but men experienced a larger drop for 2000 in comparison to 1992. The ratio of TV/TNV mobility increased among both sexes but was more pronounced for women. A look at the percentages mobile into the service class (in the lower panel of the tables) can give an explanation; this proportion increased for women but not for men.

When decomposing total vertical mobility into upward and downward moves, the trend becomes even more unfavourable for men; upward mobility (p.296) decreased and downward mobility increased for them between 1983 and 2000. Downward mobility for this period increased for women, too, but the rate for upward mobility displays a U-curve, at least between 1983 and 2000. Consequently, for 2000, the ratio of observed upward to downward mobility is more favourable for women (2.3) than for men (1.9).

This analysis is repeated for all men and women (the retired and those out of the labour force are coded in accordance with their last class position) and the results are presented in Table 12.4. The patterns for the total mobility rate or for vertical and non-vertical mobility do not seem to deviate from what we have seen before. For upward and downward mobility, women's advantage is

Table 12.4. Absolute class mobility rates in 1973, 1983, 1992, and 2000 (seven-class schema) (all males and females aged 20–69 having had a job)

1973

1983

1992

2000

Males

Total mobility rate

75.0

73.4

72.2

66.0

Total vertical

43.7

50.7

52.4

49.9

Total non-vertical

31.3

22.7

19.8

16.1

Total vertical/total non-vertical

1.4

2.2

2.6

3.1

Total upward

33.7

39.9

40.0

35.7

Total downward

10.0

10.8

12.4

14.2

Total upward/total downward

3.4

3.7

3.2

2.5

Mobility into the service class (I + II)

11.0

16.7

16.0

14.0

Mobility into the self-employed class (IVab)

1.3

1.9

4.6

7.7

Mobility into the unskilled worker class (VIIa + IIIb)

31.0

28.5

26.1

25.4

N

13,072

10,885

9,518

3,524

Females

Total mobility rate

78.2

77.5

76.1

74.4

Total vertical

37.0

47.8

50.5

53.2

Total non-vertical

41.2

29.7

25.6

21.2

Total vertical/total non-vertical

0.9

1.6

2.0

2.5

Total upward

21.4

35.4

35.5

36.3

Total downward

15.6

12.4

15.0

16.9

Total upward/total downward

1.4

2.8

2.4

2.1

Mobility into the service class (I + II)

7.4

19.1

19.7

23.1

Mobility into the self-employed class (IVab)

0.7

1.0

2.2

4.5

Mobility into the unskilled worker class (VIIa + IIIb)

32.6

32.2

33.8

31.3

N

13,429

11,581

10,580

3,676

(p.297) perhaps less pronounced in comparison to men. As the number of cases indicates, the two tables differ less for males than for females.

Unlike some of the other country chapters, we did not perform analyses in which the unemployed were classified in a separate and additional destination class. This could have been done only for the 1992 and the 2000 data because unemployment did not exist under socialism. In addition, we applied only the ‘individual’ approach and considered the location of men and women in the class structure in accordance with their own work situation. We did not recalculate our data on the basis of the ‘dominance principle’, because Hungarian families are dual earner ones in a strict sense with women working full time. Part-time employment has increased slightly in Hungary in the 1990s, but it is still rare.

Despite the fact that the total mobility rates are quite high in Hungary, the trends discussed above do not support a claim about increasing social openness. Compared to women, we can have more concern about men's mobility chances—at least on the basis of the decreasing trend of upward and the increasing trend of downward vertical mobility. This makes clear that the generally positive connotation of high mobility rates that people usually have in mind is very questionable.

Trends in relative mobility

In order to outline the pattern of social fluidity for Hungarian males and females over the last decades, the log-linear method is applied. The starting point of the investigation is a three-way mobility table: origin (O), destination (D), and time of survey (T). First, we fit a model assuming that origin and destination are independent. This is our baseline model that makes it possible to assess the extent to which further models are able to account for the total association between father's class and respondent's class. Our second model postulates unchanging relative rates as measured in terms of odds ratios. We allow variation in the origin and destination distributions at different points in time but the association between fathers' class and respondents' class is identical in the four tables, that is, we fit the CnSF to the data. The aim of applying this model is to test the validity of the FJH hypothesis for Hungary. In the third model we examine whether the odds ratios in our mobility tables differ uniformly between surveys. For this purpose, we employ the more refined Unidiff model containing both log-linear and log-multiplicative components (cf. Erikson and Goldthorpe 1992; Xie 1992). This model tests whether there were uniform changes in mobility chances in the direction of greater or less fluidity from time to time. All analyses have been carried out using LEM (Vermunt 1993).

Results of the above-mentioned models fitted to the 1973, 1983, 1992, and 2000 data are presented in Table 12.5 for men and women aged 20–69 and (p.298)

Table 12.5. Results of fitting different models to the 1973, 1983, 1992, and 2000 mobility tables (males and females aged 20–69 currently in employment or unemployed having had a job)

G 2

d.f.

Δ

rG 2

bic

Males

Seven-class schema

Ind. {OT} {DT}

5,769.6

144

15.9

4,284.7

CnSF {OT} {DT} {OD}

259.1

108

3.1

95.5

−854.6

Unidiff

202.7

105

2.6

96.5

−880.1

Unidiff parameters

1.00

0.80

0.79

0.95

(1973)

(1983)

(1992)

(2000)

Unidiff linear trend

241.7

107

3.0

95.8

−861.7

Unidiff linear trend per year

−0.0064

Eight-class schema (separating class I and II)

Ind. {OT} {DT}

5,969.9

196

16.0

3,948.8

CnSF {OT} {DT} {OD}

323.9

147

3.4

94.6

−1,192.0

Unidiff

262.7

144

2.9

95.6

−1,222.3

Unidiff parameters

1.00

0.80

0.78

0.95

(1973)

(1983)

(1992)

(2000)

Unidiff linear trend

305.4

146

3.2

94.9

−1,200.2

Unidiff linear trend per year

−0.0065

Females

Seven-class schema

Ind. {OT} {DT}

5,220.1

144

16.5

3,763.4

CnSF {OT} {DT} {OD}

197.0

108

2.9

96.2

−895.6

Unidiff

175.2

105

2.8

96.6

−886.9

Unidiff parameters

1.00

0.90

0.83

0.81

(1973)

(1983)

(1992)

(2000)

Unidiff linear trend

176.1

107

2.8

96.6

−906.3

Unidiff linear trend per year

−0.0082

Eight-class schema (separating class I and II)

Ind. {OT} {DT}

5,377.2

196

16.7

3,394.4

CnSF {OT} {DT} {OD}

254.3

147

3.3

95.3

−1,232.8

Unidiff

236.4

144

3.2

95.6

−1,220.3

Unidiff parameters

1.00

0.91

0.85

0.82

(1973)

(1983)

(1992)

(2000)

Unidiff linear trend

236.9

146

3.2

95.6

−1,240.1

Unidiff linear trend per year

20.0074

(p.299) currently in employment. The upper panel of these tables displays the results for the seven-class schema and the lower panel shows the estimates for the eight-class schema, separating the upper (I) and lower (II) service classes. It seems that the temporal invariance in all the odds ratios in the four mobility tables imposed by the CnSF model describes the Hungarian mobility regime quite well. This model captures 95 and 96 percent of the association between father's class and respondent's class, for men and women respectively. It misclassifies only about 3 percent of the cases for both sexes. The CnSF model performs fairly well, suggesting that the hypothesis of constant social fluidity between origin and destination over the last three decades in Hungary provides a good account of the data.

However, the Unidiff model, which permits the association between origin and destination to vary over time, improves on the CnSF model. This improvement is not huge; the amount of the association between father's class and respondent's class captured by this model increased only by 1 percent, as compared to the CnSF model, and the proportion of misclassified cases decreased only slightly, but the reduction in the G 2 is 56.4 and 21.8, for men and women respectively, in comparison to a loss of three degrees of freedom. For men, the Bayesian information criterion (bic) statistic also indicates that the Unidiff model is better than the CnSF model, but for women the CnSF model should be the preferred one according to bic.

Looking at the Unidiff parameters (where the value is set to 1 for 1973), the estimates reveal a particular difference between men and women in Hungary, in accordance with the above-mentioned fit statistics. For men, there was a fall in the association between origin and destination between 1973 and 1983. This trend, however, seems to have halted in 1992 and to have reversed in 2000, though social fluidity is still somewhat greater in 2000 than it used to be in 1973. Not surprisingly, fitting a linear trend to the data yields a model that is worse than the previous one by any goodness-of-fit criterion. Still, on these grounds, we can say that the annual rate in the reduction of the association between origin and destination, between 1973 and 2000, was 0.6 percent for men.

For women, however, the Unidiff parameters reveal a steady increase in the openness of society. This increase was seemingly larger between 1973 and 1983, became more moderate between 1983 and 1992 and levelled off between 1992 and 2000. When fitting the linear trend model to the data, it turns out to be the best one according to the bic statistic, though other criteria of the fit do not improve. Hungarian women experienced an annual decline in the association between origin and destination, in the period 1973 and 2000, of 0.8 percent.

Turning to the eight-class schema, as the results in Table 12.5 show, separating the upper and the lower service class does not modify the picture described above. A slight difference is that the annual rate in the increase of (p.300) social fluidity is only 0.7 percent for women in this case. Since our previous analysis on the validity of the Erikson and Goldthrope (EGP) class schema for Hungary indicated a difference between the upper and lower service class (Bukodi and Róbert 1999), we expected to get more insight into the Hungarian mobility patterns by separating these two classes. However, this assumption is not supported by our estimates.

When repeating the whole analysis for men and women aged 20–69 and including all those who are retired or out of the labour force but have been in employment earlier, our results turned out to be basically the same. The estimates presented in Table 12.6 (only for the seven-class schema) indicate a somewhat stronger annual reduction in the association between origin and destination for men, while the same reduction is lower for women—only 0.5 percent compared to 0.8 percent per year. The U-curve of the Unidiff parameters, indicating decreasing social fluidity in the 1990s, is also less marked for men.

Summarising the results of this section, we can say that Hungarian social mobility patterns have changed in the last three decades and this change indicates an increase in social fluidity between 1973 and 1983 for both sexes. Later, the increase became more moderate and for the last decade after the

Table 12.6. Results of fitting different models to the 1973, 1983, 1992, and 2000 mobility tables (seven-class schema), all individuals aged 20–69 having had a job

G 2

d.f.

Δ

rG 2

bic

Males

Ind. {OT} {DT}

7,323.9

144

16.4

5,809.2

CnSF {OT} {DT} {OD}

311.8

108

2.9

95.7

−824.2

Unidiff

262.4

105

2.5

96.4

−842.1

Unidiff parameters

1.00

0.82

0.81

0.87

(1973)

(1983)

(1992)

(2000)

Unidiff linear trend

284.7

107

2.7

96.1

−840.8

Unidiff linear trend per year

−0.0069

Females

Ind. {OT} {DT}

9,443.6

144

18.2

7,920.3

CnSF {OT} {DT} {OD}

247.1

108

2.5

97.4

−895.4

Unidiff

234.2

105

2.6

97.5

−876.5

Unidiff parameters

1.00

0.94

0.90

0.88

(1973)

(1983)

(1992)

(2000)

Unidiff linear trend

234.5

107

2.6

97.5

−897.3

Unidiff linear trend per year

−0.0048

(p.301) collapse of socialism it virtually disappeared and turned in reverse, at least for men. So, our findings do not support the claim of a secular trend toward increasing social fluidity in Hungary between the early 1970s and 2000.

Age group and cohort analysis of social fluidity

Another way to get more insight into the general trend in social fluidity is by analysing age groups. For this purpose, we used the pooled data for all the four mobility tables and we disaggregated the data to four age groups. In this case we analysed all males and females having had a job in order to gain enough cases (especially for the 2000 data with a smaller sample size). The results of the analysis according to age group are displayed in Table 12.7, for men and women, respectively.

It seems that, for all age groups of both men and women, the assumption of constant social fluidity describes the Hungarian mobility pattern quite well. The bic statistics are negative everywhere, the CnSF model captures a minimum of 84.6 percent of the association between origin and destination, and the proportion of misclassified cases is not more than 5.1 percent. Fit statistics for the Unidiff model are better only for the oldest age group. In comparison to a loss of 3 d.f., the reduction in G 2 from the youngest to the oldest age group is 6.5, 15.2, 10.0, and 44.9, respectively, for men. The same figures in the same order are 2.4, 10.1, 16.5, and 52.3 for women. It is only the oldest age group where the Unidiff model with a linear trend can be the preferred one, at least according to the bic criterion. Otherwise the Unidiff parameters display a U curve for men; social fluidity is smaller for 2000 than it used to be for 1983 and 1992. For men in the age group 30–39, the association between origin and destination is even stronger—though probably not significantly stronger—for 2000 than for 1973. This curvilinear pattern of the Unidiff parameters appears only for the youngest women. According to bic, the Unidiff model with a linear trend could be the preferred one for the two oldest age groups of women. But other parameters do not indicate much difference even for these two groups.

It seems that the age-decomposition provides even less support for any claim on a secular trend toward increasing openness in Hungarian society. Perhaps the oldest age group (50–69) could be said to display something like that: the annual rate in the reduction of the association between origin and destination was 1.8 percent per year for men and 1.1 percent per year for women. But for the younger age groups, and especially for men, fluidity remained constant over the last three decades in Hungary.

Since the analysis in this chapter relies on four cross-sectional surveys, it is obvious that some birth cohorts were sampled on more than one occasion. Consequently, it is possible to examine whether or not the strength of the (p.302)

Table 12.7. Results of fitting different models to the 1973, 1983, 1992, and 2000 mobility tables by age groups (all males and females having had a job)

G 2

d.f.

Δ

rG 2

bic

Males

Age group: 20–9 (N: 1973 = 3,307; 1983 = 2,404; 1992 = 1,704; 2000 = 767)

Ind. {OT} {DT}

1,229.8

144

12.4

68.0

CnSF {OT} {DT} {OD}

98.1

108

3.2

92.0

−875.2

Unidiff

91.6

105

2.9

92.5

−854.7

Unidiff parameters

1.00

0.84

0.86

0.90

(1973)

(1983)

(1992)

(2000)

Unidiff linear trend

95.2

107

3.1

92.2

−869.2

Unidiff linear trend per year

−0.0055

Age group: 30–9 (N: 1973 = 2,689; 1983 = 2,699; 1992 = 2,206; 2000 = 728)

Ind. {OT} {DT}

1,656.4

144

15.5

317.3

CnSF {OT} {DT} {OD}

254.1

108

5.1

84.6

−722.7

Unidiff

238.9

105

4.5

85.6

−710.8

Unidiff parameters

1.00

0.81

0.81

1.06

(1973)

(1983)

(1992)

(2000)

Unidiff linear trend

253.3

107

5.0

84.7

−714.5

Unidiff linear trend per year

−0.0029

Age group: 40–9 (N: 1973 = 2,944; 1983 = 2,234; 1992 = 2,201; 2000 = 874)

Ind. {OT} {DT}

1,616.1

144

15.8

317.3

CnSF {OT} {DT} {OD}

147.6

108

4.5

90.9

−826.4

Unidiff

137.6

105

4.4

91.5

−809.4

Unidiff parameters

1.00

0.80

0.81

0.84

(1973)

(1983)

(1992)

(2000)

Unidiff linear trend

142.9

107

4.5

91.1

−822.2

Unidiff linear trend per year

−0.0067

Age group: 50–69 (N: 1973 = 5,493; 1983 = 4,761; 1992 = 4,521; 2000 = 1,513)

Ind. {OT} {DT}

4,535.4

144

19.9

3,138.8

CnSF {OT} {DT} {OD}

228.9

108

3.7

94.9

−818.6

Unidiff

184.0

105

3.3

95.9

−834.3

Unidiff parameters

1.00

0.85

0.82

0.68

(1973)

(1983)

(1992)

(2000)

Unidiff linear trend

187.5

107

3.4

95.9

−850.2

Unidiff linear trend per year

−0.0180

Females

Age group: 20–9 (N: 1973 = 3,131; 1983 = 2,367; 1992 = 1,794; 2000 = 697)

Ind. {OT} {DT}

1,110.6

144

12.7

183.7

CnSF {OT} {DT} {OD}

86.2

108

2.6

92.2

−884.6

Unidiff

83.8

105

2.6

92.4

−860.0

Unidiff parameters

1.00

0.89

0.87

0.94

(1973)

(1983)

(1992)

(2000)

Unidiff linear trend

84.9

107

2.6

92.3

−876.8

Unidiff linear trend per year

−0.0042

Age group: 30–9 (N: 1973 = 2,736; 1983 = 2,833; 1992 = 2,392; 2000 = 728)

Ind. {OT} {DT}

1,606.1

144

15.6

−99.8

CnSF {OT} {DT} {OD}

114.9

108

3.6

92.8

−864.8

Unidiff

104.8

105

3.5

93.5

−847.7

Unidiff parameters

1.00

0.85

0.80

0.75

(1973)

(1983)

(1992)

(2000)

Unidiff linear trend

105.7

107

3.5

93.4

−864.9

Unidiff linear trend per year

−0.0096

Age group: 40–9 (N: 1973 = 3,054; 1983 = 2,411; 1992 = 2,341; 2000 = 890)

Ind. {OT} {DT}

1,805.9

144

17.0

499.6

CnSF {OT} {DT} {OD}

172.1

108

4.4

90.5

−807.7

Unidiff

155.6

105

4.2

91.4

−796.9

Unidiff parameters

1.00

0.87

0.80

0.72

(1973)

(1983)

(1992)

(2000)

Unidiff linear trend

159.0

107

4.2

91.2

−811.7

Unidiff linear trend per year

−0.0110

Age group: 50–69 (N: 1973 = 6,205; 1983 = 5,382; 1992 = 5,441; 2000 = 1,868)

Ind. {OT} {DT}

5,493.2

144

20.7

4075.1

CnSF {OT} {DT} {OD}

235.7

108

3.6

95.7

−827.8

Unidiff

183.4

105

3.1

96.7

−850.6

Unidiff parameters

1.00

0.82

0.79

0.68

(1973)

(1983)

(1992)

(2000)

Unidiff linear trend

187.9

107

3.2

96.6

−865.8

Unidiff linear trend per year

−0.0112

(p.303)

Table 12.8. Relationship between birth cohort, historical year, and age

Birth cohort

Historical year

1973

1983

1992

2000

1924–33

40–9

50–9

1934–43

30–9

40–9

49–58

1944–53

20–9

30–9

39–48

47–56

1954–63

20–9

30–9

38–47

1963–72

20–9

28–37

association between the origins and destinations of these cohorts has varied over time. This approach would give approximate information on how intragenerational life-course mobility contributes to any change in the intergenerational social fluidity of particular birth cohorts. This analysis is performed again on all males and females having had a job. Table 12.8 displays the age (p.304)

Table 12.9. Results of fitting different models to the 1973, 1983, 1992, and 2000 mobility tables by birth cohorts (all males and females having had a job)

G 2

d.f.

Δ

rG 2

bic

Males

Birth cohort: 1924–33 (N: 1973 = 2,944; 1983 = 2,097)

Ind. {OT} {DT}

1,056.2

72

17.0

442.3

CnSF {OT} {DT} {OD}

48.7

36

3.0

95.4

−258.2

Unidiff

48.7

35

3.0

95.4

−249.7

Unidiff parameters

1.00

1.004

(1973)

(1983)

Birth cohort: 1934–43 (N: 1973 = 2,689; 1983 = 2,234; 1992 = 1,775)

Ind. {OT} {DT}

1,300.3

108

16.0

348.1

CnSF {OT} {DT} {OD}

107.3

72

4.0

91.7

−527.5

Unidiff

102.8

70

3.9

92.1

−514.3

Unidiff parameters

1.00

0.87

0.87

(1973)

(1983)

(1992)

Birth cohort: 1944–53 (N: 1973 = 3,307; 1983 = 2,699; 1992 = 2,267; 2000 = 743)

Ind. {OT} {DT}

1,612.7

144

14.6

301.2

CnSF {OT} {DT} {OD}

140.0

108

3.5

91.3

−843.6

Unidiff

133.9

105

3.3

91.7

−822.4

Unidiff parameters

1.00

0.86

0.86

0.93

(1973)

(1983)

(1992)

(2000)

Birth cohort: 1954–63 (N: 1983 = 2,404; 1992 = 2,132; 2000 = 827)

Ind. {OT} {DT}

810.5

108

12.4

117.0

CnSF {OT} {DT} {OD}

62.0

72

3.0

92.3

−556.3

Unidiff

59.3

70

2.8

92.7

−541.9

Unidiff parameters

1.00

1.06

1.19

(1983)

(1992)

(2000)

Birth cohort: 1964–72 (N: 1992 = 1,550; 2000 = 697)

Ind. {OT} {DT}

404.3

72

14.7

151.5

CnSF {OT} {DT} {OD}

41.0

36

4.4

89.8

−236.9

Unidiff

39.1

35

4.0

90.3

−231.0

Unidiff parameters

1.00

1.18

(1983)

(2000)

Females

Birth cohort: 1924–33 (N: 1973 = 3,054; 1983 = 2,379)

Ind. {OT} {DT}

1,324.6

72

19.2

705.3

CnSF {OT} {DT} {OD}

60.1

36

3.7

95.5

−249.6

Unidiff

56.9

35

3.5

95.7

−244.2

Unidiff parameters

1.00

0.88

(1973)

(1983)

Birth cohort: 1934–43 (N: 1973 = 2,736; 1983 = 2,411; 1992 = 2,116)

Ind. {OT} {DT}

1,433.0

108

15.9

472.6

CnSF {OT} {DT} {OD}

94.1

72

3.9

93.4

−546.1

Unidiff

85.2

70

3.8

94.0

−537.3

Unidiff parameters

1.00

0.96

0.79

(1973)

(1983)

(1992)

Birth cohort: 1944–53 (N: 1973 = 3,131; 1983 = 2,833; 1992 = 2,421; 2000 = 808)

Ind. {OT} {DT}

1507.0

144

14.7

192.6

CnSF {OT} {DT} {OD}

86.6

108

2.9

94.2

−899.1

Unidiff

85.8

105

2.9

94.3

−872.6

Unidiff parameters

1.00

1.04

0.99

0.98

(1973)

(1983)

(1992)

(2000)

Birth cohort: 1954–63 (N: 1983 = 2,367; 1992 = 2,279; 2000 = 848)

Ind. {OT} {DT}

741.6

108

12.7

188.6

CnSF {OT} {DT} {OD}

87.6

72

3.8

88.2

−532.5

Unidiff

86.3

70

3.6

88.4

−516.6

Unidiff parameters

1.00

1.09

0.98

(1983)

(1992)

(2000)

Birth cohort: 1964–72 (N: 1992 = 1,618; 2000 = 676)

Ind. {OT} {DT}

327.0

72

13.4

230.4

CnSF {OT} {DT} {OD}

15.9

36

2.0

95.1

−262.8

Unidiff

14.8

35

2.0

95.4

−256.2

Unidiff parameters

1.00

1.16

(1992)

(2000)

(p.305) groups of five successive birth cohorts for the survey years and shows which periods are investigated when a given birth cohort is analysed. The estimates are displayed in Table 12.9 for men and women, respectively.

The results of this cohort analysis confirm what we have seen before in this chapter. The CnSF model describes the mobility pattern of these male and female birth cohorts very well. By assuming constant social fluidity over time in Hungary, our model captures nearly 90 percent or even more of the association between origin and destination and the proportion of misclassified cases is never more than 4 percent. As usual, we fit the Unidiff model to test for the possibility of uniform changes from table to table in the direction of increasing or decreasing fluidity. However, estimates indicate hardly any change for any of the male or female birth cohorts. In comparison to a loss of between one and three degrees of freedom, the reduction in G 2 is always very small and the bic values are also smaller for the Unidiff than for the CnSF (p.306) model, for all birth cohorts. Only the small decrease in the misclassified cases for the middle-aged and the younger cohorts can be considered as a sign that the assumption of uniform change cannot be fully rejected for these cohorts.

The cohort born between 1944 and 1953 is the only one represented in all four surveys we use in this chapter. For men in this cohort, the Unidiff parameters display the same U-curve we have seen before, while this reverse in the trend is not present for women. Looking at the younger birth cohorts born between 1954 and 1963, as well as between 1964 and 1972, the Unidiff parameters reveal an increase in the association between origin and destination over time. Among women this pattern appears only for the youngest birth cohort. Though the Unidiff model did not fit better than the CnSF model, we believe that it is still worthwhile to draw attention to these findings based on the Unidiff parameters because the fit of the Unidiff model to the data was not unacceptable.

Summarising what we could learn from this quasi-cohort analysis, it seems that the Hungarian mobility pattern between 1973 and 2000 can largely be described by constant social fluidity. If we keep on searching for uniform change over time in any direction, what we find is definitely not a trend toward increasing openness. On the contrary, Hungarian society seems to become less open in the second part of the period we investigated in the chapter, especially for the 1990s, after the collapse of socialism, and especially for men.

The pattern of social fluidity

So far, we have presented a global picture about changes over time in social fluidity in Hungary. In this last section we provide a more detailed description of the Hungarian fluidity pattern. By doing this we will build on the ‘core model’ of social fluidity developed in the course of the Comparative Analysis of Social Mobility in Industrial Nations (CASMIN) project (Erikson and Goldthorpe 1992: ch.4). This aimed to account for the overall pattern of association between origin and destination within the class structures of advanced societies using eight parameters (effects): two hierarchy effects (HI1 and HI2); three inheritance effects (IN1, IN2, and IN3); one sector effect (SE); and negative (AF1) and positive (AF2) affinity effect.

When fitting the core social fluidity model to the 1973 Hungarian data in The Constant Flux, the percentage of misclassified cases turned out to be relatively high. The main deviations of the Hungarian case from the core model were the following: the IN3 term defined originally for high farm inheritance had to be omitted; the positive affinity (AF2) term assuming fluidity between the petty bourgeoisie and the service class had also to be omitted; and an additional positive affinity term (AFX) had to be included indicating (p.307) mobility between farmers and agricultural workers (Erikson and Goldthorpe 1992: 151–2).

As outlined earlier, the Hungarian class structure has gradually become more similar to that of the advanced market economies. Consequently, we expected the core model to fit better for more recent periods. On the basis of the bic statistics this assumption could not be rejected but other parameters were less supportive of our hypothesis. The association between father's class and respondent's class captured by the core model turned out to be roughly the same for 1983, 1992, and 2000, for men, and for 1992 and 2000, for women. The proportion of cases misclassified by the core model was even higher for 2000. What we can definitely say is that the core model describes Hungarian mobility patterns better for the 1980s and subsequent times than it used to do for the 1970s. However, we prefer to present a Hungarian version of the core fluidity model in this chapter.

The Hungarian version of the core model has been developed on the basis of the standardised adjusted residuals calculated for the four mobility tables, for men and women separately. The design matrices are shown in appendix B (and can be compared with the matrices for the unmodified core model reported in Chapter 2). The main features of the modifications can be summarised as follows. Instead of a term defined for farm inheritance, we introduced a term for the inheritance of the service class (INX). Furthermore, we included several amendments for the positive affinity (AF2) matrix. We omitted mobility from the service class to the petty bourgeoisie. Instead of assuming a high mobility from the petty bourgeoisie to the service class for women, we expected that the daughters of self-employed fathers would tend to end up in the routine non-manual class. We also omitted the assumption of high exchange mobility between the petty bourgeoisie and farmers. We accepted the hypothesis for an intergenerational ‘upgrading’ of manual work and high mobility from the semi- and unskilled worker class to the skilled manual worker class but eliminated the reverse move between these two classes. Instead, we included two other moves, where sons of skilled worker fathers have a high probability of moving to the petty bourgeoisie class and daughters of the same fathers have a high probability of moving to the routine non-manual class. Finally, a large flow was assumed for both sexes from the agricultural worker class to the farmer class. As this move became a quite customary life course event in the 1990s when the agricultural cooperatives closed down, this can be a typical case when intragenerational mobility appeared as intergenerational class mobility. But we also kept the AFX matrix introduced by Erikson and Goldthorpe for mobility from the farmer class to the agricultural worker class.

Estimates for the fit of the Hungarian version of the core model are presented in Table 12.10, for males and females aged 20–69, currently employed or unemployed. The (p.308)

Table 12.10. Results of fitting the Hungarian version of the core social fluidity model to the 1973, 1983, 1992, and 2000 mobility tables (seven-class schema), individuals aged 20–69 currently in employment or unemployed having had a job

G 2

d.f.

Δ

rG 2

bic

Males

1973 ind.

2,558.8

36

17.9

2,223.1

core (A)

164.4

27

3.3

93.6

−87.4

1983 ind.

1,453.7

36

14.7

1,125.7

core (A)

87.3

27

2.8

94.0

−158.6

1992 ind.

1,149.3

36

14.1

829.5

core (A)

112.1

27

3.4

90.2

−128.8

core (B 5 An.s.)

111.8

29

3.3

90.3

−144.7

2000 ind.

607.8

36

17.2

324.6

core (A)

60.3

27

4.0

90.2

−153.1

core (B 5 An.s.)

59.6

30

3.9

90.2

−173.6

Females

1973 ind.

2,279.9

36

18.8

1,955.1

core (A)

270.1

27

6.3

88.1

26.5

core (B 5 An.s.)

270.4

29

6.2

88.1

9.0

1983 ind.

1,638.6

36

17.0

1,315.9

core (A)

194.0

27

5.1

88.2

−48.1

core (B 5 An.s.)

193.9

28

5.0

88.2

−57.0

1992 ind.

1,002.0

36

14.0

685.9

core (A)

61.8

27

2.8

93.8

−175.3

core (B 5 An.s.)

61.1

29

2.8

93.9

−192.5

2000 ind.

599.6

36

13.1

23.5

core (A)

52.2

27

4.9

91.2

−154.8

core (B 5 An.s.)

51.8

31

4.8

91.3

−179.9

models are fitted separately for the four tables. The parameter estimates for the core social fluidity model are presented in Table 12.11. If one parameter turned out to be insignificant, the model was re-estimated without that parameter(s). Thus, Table 12.10 includes an A and a B version of the Hungarian core model for certain tables. As the higher degrees of freedom show, the B models do not contain the non-significant parameters. Parameter estimates in Table 12.11 come always from the preferred (either A or B) models.

The Hungarian version of the core fluidity model fits the data to an acceptable degree. Even in the worst case, for the 1973 table for women, the model captures 88 percent of the association between origin and destination and the proportion of misclassified cases is slightly more than 6 percent. As far as the detailed pattern of social fluidity in Hungary is concerned, the parameter estimates in Table 12.11 reveal certain changes over time. With respect to the HI and IN effects, somewhat contradictory processes can be observed. On the (p.309)

Table 12.11. Parameter estimates of the preferred core social fluidity model (italics in Table 12.10 )a

HI1

HI2

IN1

IN2

INX

SE

AF1

AF2

AFX

Males

1973

−0.149

−0.285

0.416

1.259

0.633

−0.674

−0.507

0.427

0.801

(0.037)

(0.063)

(0.052)

(0.143)

(0.179)

(0.051)

(0.092)

(0.033)

(0.084)

1983

−0.182

−0.369

0.223

0.960

0.486

−0.622

−0.282

0.268

0.474

(0.044)

(0.060)

(0.059)

(0.146)

(0.175)

(0.056)

(0.084)

(0.036)

(0.111)

1992

n.s.

−0.497

0.384

0.593

n.s.

−0.438

−0.389

0.284

0.493

(0.065)

(0.041)

(0.104)

(0.054)

(0.101)

(0.035)

(0.145)

2000

n.s.

n.s.

0.422

0.703

0.671

−0.551

−0.583

0.307

n.s.

(0.065)

(0.211)

(0.243)

(0.087)

(0.175)

(0.059)

Females

1973

−0.123

−0.489

n.s.

1.161

n.s.

−0.906

−0.323

0.372

0.674

(0.033)

(0.067)

(0.253)

(0.046)

(0.123)

(0.033)

(0.075)

1983

−0.212

−0.600

n.s.

0.957

0.687

−0.600

−0.237

0.320

0.955

(0.034)

(0.057)

(0.109)

(0.275)

(0.055)

(0.082)

(0.031)

(0.10)

1992

n.s.

−0.542

0.217

0.615

n.s.

−0.559

−0.377

0.322

0.867

(0.060)

(0.063)

(0.095)

(0.075)

(0.093)

(0.035)

(0.173)

2000

n.s.

−0.362

0.234

0.729

n.s.

−0.898

n.s.

0.343

n.s.

(0.089)

(0.050)

(0.152)

(0.161)

(0.067)

Note: Italicised parameters are modified as shown in appendix B.

(a) Standard errors are in parentheses.

one hand, the hierarchy effects seem to disappear by 2000, at least for men, whereas for women they strengthen slightly. On the other hand, however, the inheritance effects (IN1 and IN2) display the same U-curve (a decline for 1983 and partly for 1992, and an increase for 2000) as we saw earlier in the Unidiff parameters. The pattern for the specific INX term is even more characteristic, at least for men. The service class inheritance parameter had a negative sign in 1973 and 1983 (with a decline between them), became insignificant in 1992, and turned out to be strongly positive in 2000. Consequently, the additive odds representing the propensity for immobility in the service class for males (where IN1, IN2, and IN3 apply) are 6.02 in 2000 (e0.422 + 0.703 + 0.671 = 1.796). Since the INX term is not significant for women, the same odds are only 2.61 for them. In sum, class inheritance did not increase as much for females as for males.

With respect to the sector barriers between agricultural and non-agricultural employment, the parameter estimates for SE are negative and they weaken first between 1973 and 1992 and strengthen later for 2000. This recent change, in fact, brings Hungary closer to the original assumption of the core model about stronger sector barriers in advanced societies. While sector barriers in 2000 are stronger for women than for men, the negative affinity (AF1) term, emphasising the distance between the service class and the agricultural worker class from a ‘cultural viewpoint’, is not significant for them. (p.310) For men, both terms indicate the low relative chances for mobility between the top and bottom of Hungarian society. The—strongly modified—positive affinity term (AF2) seems to be quite persistent with some temporal modifications for men. The AFX term, however, becomes insignificant for 2000. Thus, the move from the farmer class to the agricultural worker class, which was a special term for Hungary introduced by Erikson and Goldthorpe (1992), is not a characteristic type of mobility anymore, either for men or for women.

In sum, the parameters of the core model provide us with an understanding of the changing mobility regime in Hungary. The decrease in social fluidity in the 1990s largely resulted from the strengthening of class inheritance, particularly for individuals belonging to the service and self-employed classes, on the one hand, and, on the other, from the ceasing of the typical mobility routes of the earlier periods, such as mobility from farmers to agricultural labourers.

Conclusion

This chapter investigated temporal changes in Hungarian mobility patterns between 1973 and 2000. This is not the first analysis benefiting from the rich variety of Hungarian mobility data and dealing with long-term tendencies in social mobility in this country. Previously Ganzeboom et al. (1991) explored long-term changes in Hungarian social fluidity and found a gradual opening of the mobility regime. However, their research covered a period between 1930 and 1989 and did not consider the time after the collapse of socialism. Two more recent analyses also found evidence of a secular decline in the effect of social origin on social status in Hungary (Luijkx et al. 1998, 2002), but they also found some levelling off or even a slight reverse in the decreasing trend as well. These papers concluded that there was more systematic change than trendless fluctuation in the Hungarian society. Two datasets from the beginning of the 1990s (1992, 1993) have already been used in these studies, but the analytical approach applied there was based on the status attainment modelling of the mobility process. The present research as a next step in investigating long-term trends in social mobility in Hungary approached the problem from a class perspective again. Moreover, this analysis is the first one using the most recent Hungarian mobility data from 2000, which reflects the mobility experiences of a decade after the collapse of socialism.

In the chapter we explored changes in the Hungarian class structure as well as in observed mobility rates. We found that the pace of structural changes declined. The restructuring of the class distribution of Hungarian society slowed down for the 1980s as compared to its pace in the 1970s, but it increased again in the 1990s. Observed mobility rates are relatively high in (p.311) Hungary but the trend our data show does not indicate an increase in the openness of the society. On the contrary, total mobility rates decrease, especially for men. A further characteristic of the absolute mobility patterns, we want to underline, is the increase in vertical mobility—and, within this, of downward mobility. This holds true especially for men.

When turning to relative mobility rates we found that the hypothesis of constant social fluidity cannot be rejected for Hungary. We were able to detect certain trends in the mobility patterns but they did not reveal a steady increase in the openness of Hungarian society. Though an increase in social fluidity did occur in Hungary between 1973 and 1983, it levelled off later between 1983 and 1992, and reversed between 1992 and 2000. As far as the age-decomposition is concerned, only the oldest age group—50–69—showed an increase in social fluidity. For the younger ones fluidity remained constant over time. The hypothesis of constant social fluidity was supported by the analysis of birth cohorts as well, though the assumption of a uniform change for the younger cohorts could not be rejected either. For the purpose of a closer look on the pattern of social fluidity, a Hungarian version of the core model has been applied. In this respect, we want to underline the increase in inheritance effects, which also support the statement that mobility chances did not increase. On the contrary, they became worse in the last decade after the collapse of socialism. We also compared male and female mobility patterns. Focusing on the population in employment, we can conclude that the mobility chances of women did not deteriorate so much as those of men did. This is, however, a consequence of the selection effect that appears if we analyse employed women only. If we analyse all men and women, the relative advantage of women disappears.

Taking into account the main objectives of the project on the national patterns of social mobility between the 1970s and 1990s, Hungary as a case study can contribute to answering some of the research questions. Our starting point was the hypothesis about the gradual reclosure of the social structures of socialist societies (Kelley and Klein 1977, 1981). Our results support this assumption. If we divide the period of the three decades investigated here into three parts (1970s, 1980s, 1990s), the increase of mobility chances was already much more moderate by the second phase (the last phase of socialism). Thus, the hypothesis of a trend toward increased equality is only partly supported by the Hungarian case. With respect to the third phase, post-socialism brought an increase in social and especially income inequalities. The alternative hypothesis, in contrast to the assumption of an increase of equality in terms of class origin, is that intergenerational mobility chances tend to deteriorate in advanced societies. The Hungarian case seems to support this, especially if we consider our cohort analysis where the younger generations seem to be more strongly affected by the negative influences of system transformation.

(p.312) Concerning the debate on the industrialisation hypothesis (Treiman 1970), we believe that the Hungarian case demonstrates the link between industrialisation and increasing social fluidity, as our results indicate for the first phase of our time-span. However, this result cannot be interpreted in a purely liberal context. In another paper (Luijkx et al. 2002), we argue that Hungary is a country where mobility processes were influenced by political changes, which have induced institutional and economic changes toward a more modern society. Institutional changes, indeed, have been the driving force for changes in mobility chances. Both the educational system and the welfare system can be mentioned in this respect. We had no room in this chapter to analyse the role of education in intergenerational mobility (but see Simkus and Andorka 1982; Róbert 1991; Szelényi and Aschaffenburg 1993; Ganzeboom and Nieuwbeerta 1999, on the persisting influence of social origin on educational inequalities). We had no room either to provide an insight into the changes of the welfare system in Hungary, which took place in the direction of increased marketisation. We believe that these changes contributed to the decline in mobility chances that our data revealed for the 1990s. But we leave this as a task for further analyses, which need to investigate these mechanisms in a more detailed manner.

Appendix A

Data

The data for this analysis is derived from the 1973, 1983, and 1992 Social Mobility and Life-History Survey as well as the 2000 Way of Life and Time Use Survey of the Hungarian Central Statistical Office. All four national surveys have been carried out on probability samples of the Hungarian population aged fourteen and over and contain information on destination class and social origin of men and women. For this analysis we selected individuals aged 20–69 at the time of interview. The exact number of cases is shown in the tables.

Class origin is defined as the respondent's father's class when the respondent was aged fourteen. The class schema was constructed as follows: (1) detailed occupational measures were applied using four-digit (in the case of 1973 survey two-digit) Hungarian occupational codes; (2) these were then recoded into the EGP categories taking into account information on self-employment and number of subordinates.

For any comparison with other studies on Hungarian data (especially to those by Andorka 1990a, 1990b, 1997 but also Ganzeboom et al. 1991), we have to note that the EGP schema has certain deviations from the (p.313) classifications applied in those analyses, which are based on the Hungarian occupational grouping developed by Ferge and Andorka. There are two major differences between this Hungarian occupational categorisation and the EGP classes (Róbert 1998):

  1. 1. The Hungarian classification separated two groups on the top of the society: the managers (leaders) and the professionals (intelligentsia). This distinction was based on a power relationship. The equivalent classes in the EGP schema are the upper and the lower service classes (class I and II) where managers and professionals appear to be ‘mixed’ and the distinction is made on the basis of the higher or lower level of the positions.

  2. 2. Male service workers appear in Class IIIb in the EGP schema while jobs of this type in commerce, sales, transportation, communication, etc. were counted as manual and were grouped together with the skilled- or semi-/ unskilled workers by the Hungarian categorisation.

In addition to these conceptual differences, the two schemas classify certain occupations differently. An analysis investigating the EGP categorisation on recent Hungarian data of the Central Statistical Office confirmed its validity, although class IIIb (service workers) seem to be closer to class VI or in some respects to class VIIa (skilled and unskilled manual workers) than to class IIIa (routine non-manual employees) in Hungary (Bukodi and Róbert 1999). In this chapter we apply the seven (1eight) categories version of the EGP schema as presented in Table 12.A1.

Table 12.A1. Class categories, seven- and eight-class versions

Full version

Seven-class version

Eight-class version

I

Service class

Upper service class

II

Service class

Lower service class

IIIa

Routine non-manual class

Routine non-manual class

IIIb

Routine non-manual class

Routine non-manual class

IVa

Self-employed class

Self-employed class

IVb

Self-employed class

Self-employed class

IVc

Farmers

Farmers

V

Skilled workers

Skilled workers

VI

Skilled workers

Skilled workers

VIIa

Unskilled workers 1 IIIb

Unskilled workers 1 IIIb

Unskilled service workers

Unskilled service workers

VIIb

Farm workers

Farm workers

(p.314) Appendix B

Design matrices for the Hungarian modifications to the core model

1. Matrices identical to original core model and common to men and women

                   Changes in Intergenerational Class Mobility in Hungary, 1973–2000

2. Hungarian variants: men

                   Changes in Intergenerational Class Mobility in Hungary, 1973–2000

3. Hungarian variants: women

                   Changes in Intergenerational Class Mobility in Hungary, 1973–2000