Copyright Sociological Research Online, 2000

 

Jon Gubbay (2000) 'Shifting Classes: Interactions with Industry and Gender Shifts in the 1980s'
Sociological Research Online, vol. 5, no. 3, <http://www.socresonline.org.uk/5/3/gubbay.html>

To cite articles published in Sociological Research Online, please reference the above information and include paragraph numbers if necessary

Received: 23/3/2000      Accepted: 19/11/2000      Published: 31/11/2000

Abstract

Data drawn from Population Census of Great Britain suggest that both the images of proletarianisation and upward shift of the class structure are over-generalised. Shift- share analysis is used for the period 1981-1991 to explore the complex interactions between changes in class composition within industrial sectors, change in the relative size of sectors and sex composition shifts (within classes, within sectors and within class/sector categories). For example, sector shifts explain change in numbers of self-employed professionals and semi-skilled manual workers but changes in class composition within sectors account for changes in numbers of managers, non-manual ancillary workers and artists, unskilled workers and own account workers other than professionals. Change in class composition does not account for the change in the sex ratios within classes. Although sector shifts contribute to a decline of the male/female ratio in most classes, this process is uneven, with both declining male dominated and growing female dominated sectors, whose effect is partly counterbalanced by growing male dominated and declining female dominated sectors.

Keywords:
Change; Class Composition; Class Structure; Industrial Sectors; Population Census; Proletarianisation; Sex Composition; Shift-share; Social Class; Socio-economic Groups

Introduction

1.1
Braverman's (1974) sterling efforts to revise and revive proletarianisation theory have been subject to massive critical commentary, varying from attempts to refine or qualify his thesis to outright rejection of it. Consistent with Braverman's thesis, it has been argued that the growth of lower white collar occupations - especially those recruiting women - can be understood as an expansion of degraded occupations (Crompton and Jones, 1984). However, Goldthorpe scathingly challenged the supposed empirical basis for any general process of deskilling and degrading of employment. Although, he acknowledged that some case study research points to such degradation, other case studies are quite to the contrary. Anyway, this sort of evidence is '... of little relevance to the structure of employment as a whole' for it is the net effects that are at issue (Erikson & Goldthorpe 1993: 11). Indeed, Marshall and Rose (1988: 503) claim that, 'On this point, Goldthorpe seems unassailable.'

1.2
Aggregate data from the Population Censuses and large-scale social surveys do indeed appear to contradict Braverman since they indicate continuing relative decline of all forms of manual employment and growth of managerial, professional and other white collar jobs. Even Wright's initial defence of proletarianisation theory took a restricted form, namely that there was continuing proletarianisation within industrial sectors in the 1960s. Proletarianisation had been masked at the all-sectors level by the tendency for relatively slow growth or decline of those sectors which had the highest proportion of proletarians but he argued that it was revealed within sectors by the use of shift-share analysis. (Wright & Singelmann, 1982) As it turned out, Wright's subsequent updating of the research using similar techniques led him to retract even his restricted proletarianisation theory, thus supporting Goldthorpe's position, since his later studies showed upgrading within sectors as well as upgrading due to the differential change in sector sizes (Wright & Martin, 1987; Wright, 1997: Ch. 3).

1.3
This paper is not so much an attempt to give comfort to any of the protagonists of these debates as to insist that the relative importance of within- sector and differential sector shift factors varies according to sex and class. On the basis of the data presented here, it is argued that one cannot collapse complex patterns of change in occupational structure into the exclusive alternatives of establishment or refutation of proletarianisation. Caution is especially important given the lack of definitive evidence as to the extent of slippage over time between occupational titles and the content of occupations (Crompton 1980; Goldthorpe 1980). Accordingly, apart from providing a convenient resource for students and researchers, this paper has modest objectives in exploring variability in changes in class composition according to sex and sector. For example, the analysis will address the following questions:

Changes in Economic Activity and Employment 1961-1991

2.1
Between 1971 and 1991, while there has been an overall increase in the economic activity rate, it has declined for men but increased for women. (See Table 1) These changes are quite strongly age- related, with the proportion of older men who are economically active declining markedly. The rate for the 55-59 age range declined from 95.4% to 80.2% and, in the age range 60-64, it declined from 86.7% to 57.0%. (Click here for Chart1 Men.) There has also been a significant increase in the proportion of economically active women but this has been due to growth in the proportion within the 20-59 age range (i.e. from 54.6% to 69.0%) rather than at older ages. (Click here for Chart 1 Women.) There has been little change in economic activity rates among men and women aged less than 20.

2.2
Broadly comparable statistics for the composition of the economically active population are available for the Population Censuses from 1961 through to 1991 but they differ markedly in respect of the proportion who were unemployed (or on a Government scheme). In the 20-64 age range, the percentages of the economically active persons who were unemployed or on a Government Scheme were 2.7%, 4.7%, 8.9% and 9.5% in the successive census years. (The percentages were very much higher among those aged less than 20 in 1981 and 1991, 18.1% and 27.1% respectively.) However, the concern of this paper is the composition of places within the sector/class structure rather than the economic status of persons, so the following analysis focuses on those who are 'in employment'. Accordingly, Table 2 presents a breakdown of the persons 'in employment' across all ages. The following sex difference and over-time trends are apparent:

2.3
Fuller details of percentages of those 'in employment' broken down into age categories are provided in Table 3, as drawn from the published 10% samples in the Population Censuses. These results permit some elaboration of the above points, as follows:

2.4
The broad picture of change that emerges from the data considered thus far is of the development of high levels of unemployment, growth of self-employment, growth of female employment (including part- time employment) and decline of male employment (particularly in the older age groups). More recently there has been a significant growth in both male and female part-time employment of young people.

Class Shifts 1961-1991

3.1
On the face of it, the Population Census data of Socio-Economic Groups (SEG) and Social Class (SC) since 1961 give strong support to the view that there has been a general upgrading of the class structure. However, in assessing this evidence one needs to be aware of considerable hazards in the measurement of change in class composition based upon the Population Censuses.

3.2
First, neither classification schema is derived from well-articulated theoretical principles. Rose and O'Reilly (1997: 4) cite research which rejects claims that SC accurately measures either 'general standing in the community' or 'levels of occupational skill' - and it is notable that the switch between these two quite different criteria from 1970 to 1980 was accompanied by little change in allocation of occupations (Reid 1989: 53-4). SEG, which originates from 1951, is an improvement over SC in that it seeks to distinguish employees from self-employed persons and professionals from managers but, as Rose and O'Reilly (1997: 6) note , '... nowhere can one find an explanation of the conceptual basis of SEGs.'.

3.3
Second, over- time shifts in the class allocation of particular occupational titles (or occupation/economic status combinations) are difficult to trace and estimate. Sometimes work of similar content is re-titled and perhaps reallocated whereas in other cases the content of work changes drastically without corresponding change in occupational title.

3.4
Third, the bases for the published statistics has changed from one Census to another, for example with regard to geography (Great Britain or England and Wales), numbers who are economically active or just those 'in employment', aged over 15 or over 16 etc. Furthermore, most of the published tables do not distinguish between full-time and part-time employment.

3.5
It is not the purpose here to investigate these issues but merely to note them as a caution to readers. Given that the Census bureau makes considerable efforts to preserve continuity of classification, such objections to the data may not serious where only a relatively short period of time is under investigation since the there will then be only limited amount of slippage between occupational titles and content. Accordingly, apart from the next paragraph, the remainder of this paper is limited to consideration of one inter-censal period.

3.6
It is, then, with some trepidation that Table 4 is presented here to show, over thirty years, the changing proportions in each SEG of men 'in employment'. (Click here for a chart summary.) Note that there are no comparable statistics for this series with regard to women.

3.7
Salient points to note include the following:

3.8
The trends identifiable between 1961 and 1991 can be explored more fully for the last decade of this period in respect of sex, sector and class. Unfortunately, comparable data is not available for the earlier censuses. The remaining sections of this article analyse, for Great Britain in 1981 and 1991, the cross-tabulations of the SEG and SC class schemas with sex and sector. The 1991 cross-tabulation of sector with SEG was specially commissioned and appears here for the first time. The other cross-tabulations appear in the printed Population Census reports on economic activity. Click here to access the datasets.

3.9
Although there is a high degree of comparability between 1981 and 1991 in the criteria for allocating respondents to occupations, some adjustments have been made in order to take account of known changes in assignments of occupations to classes. Appendices 1A, and 1B explain the procedure adopted to effect the 'correction' for SC and SEG respectively. All the following analysis is based on these 'corrected' datasets.

Class and sector shifts 1981-1991

4.1
In this section, consideration is given to the extent to which, regardless of variations between the sexes, shifts in the relative size of sectors can account for the changes in the class composition. An explanation of shift-share analysis is provided in Appendix 2A, where it is shown how overall change in class composition can be decomposed into three elements - effect of differential change in sector size (sector shift effect), effect of changes in class composition within sectors (class composition change effect) and effect of overall growth in persons 'in employment' (uniform change effect).[1]

4.2
Accordingly, a shift-share analysis for SCs is presented in Table 5 and summarised as a chart. This appears to be a rather daunting table but, in fact, its interpretation is quite intuitive. This will be illustrated by reference to the statistics for 'Professional etc. occupations'. The number of professionals in the sample increased over the decade from 92621 to 114602, that is by 21981 (an increase of 23.7% of the number of professionals in 1981). This increase is composed of the three elements indicated above:

  1. Sector shift effect (i.e. 11147, which is 12.0% of the number of professionals in 1981),
  2. Class composition change effect (i.e. 8667, which is 9.4% of the number of professionals in 1981), and
  3. Uniform shift effect (i.e. 2167, which is 2.3% of the number of professionals in 1981).

4.3
Examination of Table 5 indicates that the impact of change in industrial structure on the relative size of social classes varies considerably from one class to another.

.

4.4
SEG composition varies considerably from sector to sector and according to sex. For 1981, calculations have been made to assess which SEGs are concentrated in particular sectors. Some of the results are entirely to be expected, in particular that farmers and agricultural workers are concentrated in the agriculture and horticulture sector and that the armed forces are concentrated in the public administration sector. Other calculations, although not perhaps surprising, are nevertheless of considerable interest and the actual percentages are occasionally quite striking.

4.5
The impact of sector shifts on SEGs is now examined by means of shift-share analysis, as presented in Table 6 and summarised as a chart. The results confirm and elaborate the conclusions reached above on the basis of shift-share analysis of sectors and SCs. Again, we see variation from one class to another in the extent to which changes are accounted for by sector shift and class composition change effects.

.

Class and Gender Shifts 1981-1991

5.1
Table 7 presents the SC distribution of the 10% samples from the Censuses by sex. (Click here for a graphical summary of the data - Chart 7) From this table one can calculate that the number 'in employment' grew by 2.3% - but this was made up of a 12.8% increase in women and a 4.6% decline in men. As is well known, there are considerable disparities in the proportions of men and women in the various classes. Males are greatly over- represented among professionals and skilled manual workers. They are also somewhat over- represented among intermediate and partly skilled manual occupations. In contrast, women are greatly over-represented in skilled non-manual occupations and somewhat over-represented in unskilled manual occupations.

5.2
Over the decade, there have been significant changes in SC composition and gender shifts within these categories. There was very substantial growth in numbers of both male and female professional and intermediate occupations, although in proportionate terms female growth was more rapid than male growth. In contrast, the predominance of women in the skilled non-manual SC in 1981 became even greater by 1991.

5.3
The shift-share analysis presented in Table 8 enables a significant conclusion to be reached about the aggregate effect on gender composition of changes in the relative size of SCs. It was not attributable to relatively high growth of the SCs where women predominate and relatively low growth or decline in those where males predominate. Indeed, the overall male/female ratio declined from 1.50 in 1981 to 1.27 in 1991 but virtually that entire decline - to 1.28 - is accounted for by changes in proportions of men and women within SCs.

5.4
Table 9 and sets out the distribution of SEGs from the 10% Census samples and the associated chart indicates how they grew or declined between 1981 and 1991. Table 10 has been calculated to show the male/female ratios at 1981 and 1991 - and these are also indicated in the associated chart. These results permit confirmation and elaboration of conclusions reached above with regard to SC. Men are massively over-represented among professional workers (both employed and self-employed), skilled manual workers, manual foremen and supervisors, farmers and members of the armed forces. They are also considerably over-represented among employers and managers, non-professional own account workers and agricultural workers. Women are considerably over-represented among personal service workers and junior non-manual workers.

5.5
Turning now to the changes in the relative size of SEGs and the sex shifts within them, the following points can be readily noted:

5.6
Given that the number of women 'in employment' grew by 12.8% while declining by 4.6% for men, this factor in itself can be expected to have contributed to the reduction in the proportion of males within SEGs. However, in some SEGs the proportionate change in males was much less than can be accounted for by the changes in total numbers of men and women (e.g. skilled and unskilled manual workers, personal service workers and own account workers other than professionals) - and in some cases much more (e.g. managers, professionals and foremen and supervisors).

5.7
The shift-share analysis set out in Table 11 decomposes differential change in size of the SEGs, the sex shifts within SEGs and the overall increase in numbers 'in employment'. It confirms the conclusion reached for SCs above in showing fairly decisively that changes in class composition have only a very small effect on the decline in the male/female ratio (i.e. from 1.50 to 1.46). Rather, the major factor is the sex shift within the SEGs (i.e. reducing the male/female ratio from 1.50 to 1.30).

Sector and Gender Shifts 1981-1991

6.1
Sector data drawn from the 10% Census samples show in Table 12 the male/female ratios within sectors and the percentage changes in sectors sizes between 1981 and 1991. The most male dominated sectors include coal extraction, construction, transport, engineering and vehicle and other metal manufacturing while women are considerably over- represented in health, personal, educational and other services to the general public and footwear and clothing industries. The table well indicates the sector shifts that are often summarised as 'de- industrialisation'. Two large industrial sectors - coal extraction and metal manufacture declined by more than half while a whole range of service sectors increased substantially. Other sectors declining substantially included vehicle and other metal goods manufacture; engineering and chemicals; railways; electricity and gas production; food, drink and tobacco manufacturing and textiles, footwear and clothing. In short, there is the familiar picture of decline in employment in manufacturing and extraction with only the rare exception like gas and oil. Most areas of service employment grew significantly, although rather modestly in the cases of education, post and telecommunications. The sectors of growing employment included business, financial, sanitary and cultural services.

6.2
There have been particularly dramatic cases of male dominated sectors declining drastically in size, including coal extraction, metal and vehicle manufacturing. On the other hand, there were also declining female dominated sectors such as footwear and clothing industries. Moreover, there was considerable growth in numbers employed in female dominated medical and health services, hotels and catering - although it is true that there was greater proportional male than female growth. Banking and finance and other services provided to the general public grew strongly, with women increasing their already significant numerical preponderance. Accordingly, one should be cautious about generalising too strongly from extraction and heavy industries towards the proposition that sectors where the male/female ratio was high tended to reduce their male employment more than those where it was low. This issue can be satisfactorily assessed by means of a shift-share analysis.

6.3
As shown in Table 13, the overall decline in the male/female ratio is accounted for both by the combined effect of the sex shift within sectors and sector shift effect, the former factor playing a somewhat larger role. A summary way of seeing this is to note that the male/female ratio falls from 1.50 to 1.36 just as a result of differential change in the relative size of sectors but that the effect of the within-sector change alone is to reduce the ratio by a somewhat smaller amount, to 1.40.

Class, Sector and Gender Shifts 1981- 1991

7.1
In this final substantive section, shift-share analysis is extended along the lines propounded by Smith (1991) to deal with changes in relative proportions of the two sexes within sector/class categories as well as overall within sectors taken as a whole and classes taken as a whole. The detailed rationale for the corresponding shift-share analysis, following Smith, is set out in Appendix 2B. The results for the SC class schema are set out as Table 14. This is quite a complex table showing varying patterns of change both by sex and SC. Nevertheless, the key findings can be stated quite straightforwardly as follows:

7.2
It is not possible to carry out a similarly full shift- share analysis for the sex shifts within sector/SEG categories. This is because the calculation involves ratios of numbers of men (and women) in sector/SEG categories but many of these numbers are zero or quite small - and, accordingly, the ratios would be meaningless or likely to distort the estimated effects. A partial shift-share analysis not decomposing the class composition change and the sex shift effects will be discussed shortly but, first, a rather more roundabout approach is adopted in order to gain some grasp of the effects of sex shift within SEG/sector categories.

7.3
By excluding low value cells from the analysis, estimates were constructed of the effects produced by changes in the sex ratio within sector/SEG categories. The procedure used was to construct a matrix where non- excluded cells represented the expected numbers of men on the assumption that their numbers changed in the same proportion as did the number of 'all persons' in that sector/SEG. These numbers were then subtracted from the corresponding number of men in that sector/SEG in 1991 in order to estimate the extent of growth or decline in numbers of men in a SEG/sector category not accounted for by the growth or decline of the SEG/sector category as a whole. To estimate the overall effects on sectors and classes of changes in the sex ratios within SEG/sector categories, the row and column totals have been calculated. (Note that it is not necessary to carry out the corresponding calculations for women since they are simply the negative of those for men.) The results are set out in Table 15.

7.4
For example, there were 24264 managers in small establishments employed in the Retail sector in 1981 and 31053 in 1991, that is a growth of 28%. If the number of men in this SEG employed in the Retail sector had grown in proportion, the expected number would have risen from 16313 in 1981 by 28% to 20878 in 1991. In fact, in 1991 there were 18450 male managers in small establishments in the retail sector. Accordingly, it can be argued that the sex shift against men and in favour of women managers in small establishments within the retail sector accounted for an increase in their numbers in this SEG/sector category of 2428 (i.e. 20878-18450). Other notable sex shifts against male employment in this SEG occurred in Business Services (3508) and wholesale distribution (1235). The total effect of the sex shift for managers in small establishments can then be calculated by summing similar figures for that SEG across all sectors. As shown in Table 15, this total is -15536 which is, of course, the negative of the corresponding total for women. Comparing this figure with 92475, the total number of male managers in small establishments, it can be estimated that the sex shift effect against men in this SEG is 16.8%. This is a substantially greater decline than the overall percentage decline in men in employment, 4.6%. (For women in this SEG the sex shift effect is 53.5%, which greatly exceeds the proportionate increase in female employment of 12.8%.) Accordingly, the conclusion can be drawn that the growth in proportion of females in this SEG is due to a substantial extent to the overall decline of the male/female ratio within the sectors of the SEG, not just overall growth in numbers of females in employment or a change in the relative size of sectors.

7.5
Table 16 is derived from the column totals of Table 15. It shows the differences between actual and expected numbers in proportion to male and female SEG sizes in 1981. That is, it indicates the effect on SEGs of sex shifts within the SEG/sector categories. Leaving aside the small SEGs, it shows significant female growth in employment at the expense of men within SEG/sector categories for managers in large establishments, employed professionals and foremen and supervisors of manual workers. There were a number of substantial SEGs within which there was less sex shift effect to the advantage of women, indeed less than the overall growth of females in employment; these were junior non-manual workers, personal service workers, skilled, semi-skilled and unskilled workers).

7.6
Similarly drawing on the row totals of Table 15, Table 17 shows the differences between actual and expected numbers in proportion to male and female sector sizes in 1981. It is noticeable that there was a strong male to female shift in a number of substantial sectors including sanitary services, business services, manufacture of office machinery, supporting services to air transport, miscellaneous transport services, owning and dealing in real estate and research and development. In a range of others, there was little or no female increment beyond expected numbers - for example in retail, hotels and catering, footwear and clothing, textiles, food, drink and tobacco, education and health services - and well less than the percentage increase in female employment overall.

7.7
Now we turn to consideration of the partial shift- share analysis, as shown in Table 18. Although this is a rather large table, it is straightforward to identify some of its most significant findings when read in conjunction with Table 17, as follows:

Conclusions

8.1
This study demonstrates that neither changes in the relative size of classes nor the sex composition of them are reducible to differential size of sectors and the relative changes in numbers of men and women in employment. There are also changes in class composition within sectors and sex composition within sector/class categories which impact to varying extents from class to class.

8.2
There has been a decline in unskilled and manual employment and growth in managerial and professional employment and in self- employment but these changes are explicable in terms of sector shifts in only some classes. Sector shifts do explain growth and decline of self-employed professionals and semi-skilled manual workers respectively. However, changes in class composition within sectors must also be invoked to explain changes in numbers of employed professionals, foremen and supervisors of manual workers and skilled manual workers. Indeed, class composition change effects are the main or exclusive factor in changing numbers of managers, non-manual ancillary workers and artists, unskilled workers and own account workers other than professionals. A particularly interesting case is junior non-manual workers where an increase promoted by sector shift was cancelled by a decrease due to class composition change.

8.3
There has been a growth of female employment and decline of male employment but it has had a far from even effect on sex ratios within classes. It has been shown that changes in class composition within sectors did occur to a substantial extent but that this does not account for the changes in the sex ratios within classes. Overall, sector shifts do contribute to a decline of the male/female ratio in most classes but this process is quite uneven, with both declining male dominated and growing female dominated sectors, whose effect is partly counterbalanced by growing male dominated and declining female dominated sectors. In any case, there are variations within sectors in the extent of change in the sex ratio, again with differential impact on sex ratio of the classes. Finally, it has been shown that numbers of males and females experienced different patterns of growth and decline in classes accountable to impacts, varying according to class, of uniform change, sector shift, class composition change and sex shift within sector/class categories. The sex shift effect was particularly important for employers, managers, professionals and supervisors but it was complemented to different degrees by sector and class composition change effects. There were some SEGs where the effects interacted in particularly complex ways. For example, sector shifts tending to increase the number of junior non-manual workers (particularly men), class composition changes to reduce them and sex shift to bring about the result of male decline and female increase.

8.4
Shift-share analysis has proved to be a useful and convenient means of disentangling composite changes, in an accounting sense, among several factors in large datasets. However, an approach using a pair of snapshot distributions cannot identify causal processes, still less how they impact on individuals. To explore those sorts of concerns, the type of analysis presented here needs to be complemented by case studies in particular sectors and analysis of panel data.

Notes

1. Appendix 2A also explains the algebraic notation used in tables 5, 7, 8, 11, 13, 14 and 18.

Acknowledgements

I am very grateful to the Office of National Statistics for permission to use data from the published Population Censuses and the specially commissioned table. In particular, I would wish to thank Linda Alderslade for her efficiency and helpfulness in producing this table. I should make it quite clear that I take full responsibility for any errors of conceptualisation or execution in deriving the 'corrected' versions of SEC and SEG as explained in Appendices 1A and B.

References

BRAVERMAN, H. (1974) Labor and Monopoly Capital New York: Monthly Review Press.

CROMPTON, R. (1980) 'Class Mobility in Britain' Sociology 14: 117-9.

CROMPTON, R. & JONES, G. (1984) White Collar Proletariat: Deskilling and Gender in Clerical Work London: Macmillan.

ERIKSON, R. & GOLDTHORPE (1993) The Constant Flux Oxford: OUP.

GOLDTHORPE, J. (1980) 'Class Mobility in Britain: A Reply to Crompton' Sociology 14: 121-3.

MARSHALL, G. & ROSE, D. (1988) 'Proletarianization in the British Class Structure?' British Journal of Sociology 39: 498-518.

OPCS and Department of Employment Group (1991) Standard Occupational Classification, Volume 3.

OPCS Census of POPULATION - volumes as cited accompanying tables.

REID, I. (1989) Social Class Differences in Britain 3rd edition Glasgow: Fontana.

ROSE, D. & O'REILLY. (eds.) (1997) Constructing Classes London: ONS and ESRC.

SMITH, S.A. (1991) 'Shift Share Analysis of Change in Occupational Sex Composition' Social Science Research, 20: 437-453.

WRIGHT, E. O. & SINGELMANN, J. (1982) 'Proletarianization in the Changing American Class Structure' American Journal of Sociology, 88 (Supplement) S177- S209.

WRIGHT, E. O. & MARTIN, B. (1987) "The Transformation of the American Class Structure, 1960-1980" American Journal of Sociology, 93: 1, 1-29.

Copyright Sociological Research Online, 2000