Jump to ContentJump to Main Navigation
North-South Trade, Employment and InequalityChanging Fortunes in a Skill-Driven World$

Adrian Wood

Print publication date: 1995

Print ISBN-13: 9780198290155

Published to Oxford Scholarship Online: November 2003

DOI: 10.1093/0198290152.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2017. 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 http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 16 January 2017

(p.435) Appendix A4 Occupational Relativities

(p.435) Appendix A4 Occupational Relativities

Source:
North-South Trade, Employment and Inequality
Publisher:
Oxford University Press

This appendix continues the review of recent trends in skill differentials in the North (started in Appendix A3), using occupation as the indicator of skill. The first section looks at the relative wages of white‐collar and blue‐collar workers, the second at more finely specified occupational wage relativities. The third and fourth sections examine movements in relative unemployment and vacancy rates, and in employment levels. The last section documents the sources of data used.

Although detailed occupational classifications are concerned mainly with differences in the nature of work (nurses versus librarians, plumbers versus electricians), some of the broad categories involved can be used as indicators of skill, because they are associated with differences in required amounts of education, training, and experience—for example, the distinctions between professional and clerical workers, or craftsmen and labourers.1 Moreover, occupational data have a potential advantage over educational measures of skill, which is that they may also capture training undertaken outside the formal schooling system, such as apprenticeships.

It is important to recognise, however, that occupational categories are at best a proxy for differences in education and training. This matters in the present context because the required amounts of skill in particular occupations are likely to alter over time, which makes changes in wage relativities among occupations an inaccurate guide to movements in the rewards to different levels of education and training. For example, the wage gap between two occupations could widen or narrow even in the absence of change in true skill differentials if the amount of skill needed in one or other occupation were to alter. Similarly, a rise in the relative skill content of lower‐paid occupations might conceal an increase in the wage differential between skilled and unskilled workers.

In addition to this limitation of principle, there are some troublesome practical problems in using occupational wage data to measure changes in skill differentials. It is rarely possible to control for differences and changes in age structure—for example, fast growth of employment has tended to reduce the average age and experience of professional workers, making their average earnings growth slower than it would appear if data for specific age groups were available. Sometimes it is not even possible to distinguish males from females, so that movements in the relative wages of different occupations may also be affected by differences and changes in the relative numbers of men and women employed, or in their relative wages. Moreover, occupational categories are often coarse and changing mixtures of activities with widely varying skill requirements and wages.

(p.436) A4.1 Relative Wages of White‐Collar and Blue‐Collar Workers

The coarsest of all occupational breakdowns is that between white‐collar (non‐manual) and blue‐collar (manual) workers. Its particular weakness in the present context is that the white‐collar and the blue‐collar groups are both mixtures of SKILD and BAS‐ED labour (see Section A1.5). On average, non‐manual workers have more education and training than manual workers, but many blue‐collar jobs require more training than many less‐skilled white‐collar jobs. For about half the countries in the North, however, this is the only breakdown of wages by occupation for which time series are available.

The trends in the relative wages of white‐collar and blue‐collar workers in fifteen Northern countries are summarised in three graphs, whose design and data owe much to earlier work in OECD‐EO (1987, ch. 3).2 In most cases (and unless otherwise mentioned), the data refer only to males and only to manufacturing or ‘industry’, but the coverage of the white‐ and blue‐collar categories varies among countries. Moreover, in these and in subsequent graphs, as detailed in the notes, some of the lines have been shifted upwards or downwards to improve readability. At best, then, these figures give only a rough indication of movements over time.

Fig. A4.1 refers to relative wage levels in five of the largest countries: for France, the UK, and the USA, the data are not confined to industry. In three of the five countries (Germany, the UK, and the USA), white‐collar wages rose faster than blue‐collar wages after the late 1970s, which suggests a widening of skill differentials. In Germany, there was such a rise also in the early 1970s, but from 1975 to 1979, and in the UK and the USA during most of the 1970s, the white‐to‐blue‐collar wage ratio was constant or declining. In Japan, this ratio declined from the early 1960s to the mid‐1970s, after which it changed very little. In France, there was a steep decline from the late 1960s, which slowed down but did not altogether cease in the 1980s. However, part of the downward trend in France was due to changes in age structure and ‘classification drift’ among managers (Marsden 1989b: 1.4, 2.5; Cases and Lollivier 1989).

The next figure needs to be interpreted more cautiously. It does not show relative wage levels, but ratios of wage indices for white‐ and blue‐collar workers. Moreover, in most cases (and unless otherwise mentioned) the two indices of which the ratio is shown here are based on different payment periods, usually hourly wages for blue‐collar workers and monthly salaries for white‐collar workers. Thus if weekly or monthly hours of work were to fall over time, as they did in most countries during part or all of this period, this ratio of indices would decline even if there were no change in the true ratio of (hourly or monthly) (p.437)

Appendix A4 Occupational Relativities

Fig. A4.1 Relative Wages Of White‐Collar and Blue‐Collar Workers

Notes and sources: Cross‐country differences in the level of wage differentials are not accurately reflected in this graph, which is intended only to illustrate trends over time. The coverage of the data varies across countries. In addition, to make the graph easier to read, 0.05 has been added to the figures for Germany, 0.45 to those for Japan, and 0.15 to those for the USA. For more detailed definitions and sources of data, see Sect. A4.5.

white‐collar to blue‐collar wages. To get a better idea of what has really been happening to this wage ratio, the graphs should be mentally tilted anti‐clockwise (OECD‐EO 1987, n. 15).

Fig. A4.2(a) covers two large countries (Canada and Italy) and three smaller ones. For Canada, Denmark, Finland, and Sweden (after 1980), the sectoral coverage extends beyond industry. In Denmark, the first few years include both males and females, as do all the data for Finland and (probably) Italy: this may impart a downward bias, because of the increasing share of (lower‐paid) females in the white‐collar group.3 In all five countries, the white‐collar index rose more slowly than the blue‐collar index during most of the 1970s, although in three of them (Canada and Italy after 1974 being the exceptions) this probably partly reflects a fall in hours of work. In Sweden after 1980, working hours have no further effect (because of a change in the source of the data), but the white‐collar to blue‐collar wage ratio continued to decline. In the other four countries, however, the earlier trend ceased or was reversed. In Finland, the change occurred soonest (1975), but the subsequent upward tendency was only slight. In Canada and Denmark, there was an upturn after about 1980, although in both cases with a drop in 1986–8. (p.438)

Appendix A4 Occupational Relativities

Fig. A4.2 Changes In White‐Collar Relative To Blue‐Collar Wages (Ratio Of Monthly Salary To Hourly Wage Indices Unless Otherwise Indicated)

Notes and sources: see previous figure. To improve the readability of the graph, 4 has been subtracted from the Canadian index and 14 from the Swedish index.

Notes and sources: See Fig. A4.1. To improve the readability of the graph, 12 has been added to the index for Austria and 7 to the index for Norway, while 20 has been subtracted from the index for Belgium, 2 from the index for The Netherlands, and 5 from the index for Switzerland.

(p.439) And in 1983 the relative wage of white‐collar workers began to rise in Italian industry.4

Fig. A4.2(b) covers a further five small countries: in The Netherlands (after 1980) and Switzerland, females as well as males are included. In all five countries, the ratio of white‐collar to blue‐collar wage indices tends to decline in the earlier part of the period, even in Austria and The Netherlands, where the series are not affected by changes in working hours. In the latter part of the period, although the general downward tendency ceases, the pattern is mixed. Only in Austria is there a sustained subsequent upward trend (from 1975). In Norway, there is a marked upward movement from 1979 to 1986, but then a decline to 1989. In The Netherlands and Switzerland, there is little change after the mid‐1970s. And in Belgium, the earlier decline continues during the 1980s.

In summary, the data on white‐ and blue‐collar wages suggest that skill differentials in many Northern countries widened after the late 1970s or early 1980s, though not as consistently as the data on wages by education and age.5 In five of the fifteen countries, the graphs show clear and sustained increases (Austria, Germany, Italy, UK, USA). In seven countries, there was a less clear uptrend, or merely cessation of a previous downward tendency, but in four of these (Denmark, Finland, Norway, Switzerland), the indices may be biased downward by reductions in working hours. Only in three countries (Belgium, France, Sweden) did the downward tendency continue in the 1980s. Moreover, these series may understate or conceal widening of skill differentials because of the lack of control for age (and in a few cases also for sex) mentioned earlier.

A4.2 More Specific Occupational Wage Relativities

Looking more closely at the blue‐collar category, Fig. A4.3 presents evidence for six countries on trends in wage differentials between skilled and unskilled manual workers (variously defined). The data refer to males, except in Denmark, where the skilled group includes a few females, in Italy, where sex is not specified, and in the USA, where both sexes are included from 1970 onwards. For Denmark, Germany, and Italy, the coverage is limited to industry, but elsewhere other sectors are also included.

The figure suggests striking differences among countries. In both the UK and the USA, there is a steep increase in the skilled‐to‐unskilled manual wage ratio after the late 1970s. In the USA (where the rise in the 1980s is similar in the male‐only data in Table A4.3), this continued a trend that began in the early 1970s. In the UK, by contrast, the increase merely reversed a steep decline during (p.440)

Appendix A4 Occupational Relativities

Fig. A4.3 Relative Wages Of Skilled and Unskilled Blue‐Collar Workers

Notes and sources: See notes to Fig. A4.1 and Sect. A4.5. To improve the readability of the figure, 0.20 has been added to the ratio for France, 0.05 to the ratio for Italy, and 0.17 to the ratio for the USA, while 0.06 has been subtracted from the ratio for Germany.

the earlier part of the 1970s, when incomes policies discriminated in favour of lower‐paid workers, so that in 1990 the skilled‐unskilled wage ratio was no higher than it had been in 1970.6 In France and Italy, this ratio declined for most of the period for which data are available, with a particularly steep decline in Italy during the 1970s (a pattern also observed in Sweden 1975–80).7 But in both countries it eventually levelled off, and may even have risen in the 1980s in Italy.8 In Germany and Denmark, there was much less change, though the German data suggest a slight decline from the mid‐1970s to the late 1980s. In Austria in the 1980s, too, there was little change in the relative pay of skilled and unskilled manual workers.9

The pay of lower‐level white‐collar (mainly clerical) workers is generally similar to that of blue‐collar workers. During the 1970s, moreover, there was little change in the wage ratio between these two groups (OECD‐EO 1987, table 3.2 and chart 3.3). Relative to blue‐collar workers, lower white‐collar workers became slightly (p.441) worse off in France and Italy, slightly better off in Germany and the USA, and held their ground in the UK after 1973. During the 1980s, too, as shown in Table A4.1, there were only small movements in this wage ratio, and not in any consistent direction. In both decades, therefore, the movements in the average white‐collar to blue‐collar pay ratio charted in the previous section mainly reflected the changing relative position of higher‐level (professional, technical, and managerial) white‐collar workers.

Table A4.1. Relative Wages of Lower White‐Collar and Middling Blue‐Collar Workers in 1980s (Ratio of Male White‐Collar to Blue‐Collar)

Early ratio

Late ratio

Late/early

Austria

1983–9

1.11

1.09

0.99

Canada

1981–8

0.96

0.95

0.99

France

1980–5

1.24

1.22

0.98

1987–90

1.15

1.15

1.00

Germany

1980–9

1.22

1.22

1.00

UK

1980/1–89/90

0.92

0.94

1.02

USA

1984–90

1.18

1.15

0.97

Notes and sources: sources as in Sect. A4.5 except where otherwise specified below.

Austria: Angestellte mit gelernter Tätigkeit versus Angelernte Arbeiter.

Canada: clerical versus non‐farm manual workers.

France: 1980–85, employés versus ouvriers specialisés in private and semi‐public sectors; 1987–89, employés versus ouvriers non‐qualifiés in private sector.

Germany: Kaufmannische Angestellte Leistungsgruppe III versus Arbeiter Leistungsgruppe II in industry (including trade and finance for Angestellte). Arbeiter × 4.34 for weekly/monthly wage comparability.

UK: clerical workers (group VII) versus processing, making, and repairing workers, mechanical and engineering (group XIV).

USA: administrative support workers versus machine operators: source as in Table A4.3.

The next figure looks more closely at relativities within the white‐collar category. Except for the USA, where females are also included after 1970, the data refer only to males, and cover quite a range of sectors, except in Norway and Sweden, where they refer to manufacturing. In three of the four large countries in Fig. A4.4(a), the wage differential between higher and lower white‐collar workers widened during the 1980s. This movement is more marked in the UK and the USA (see also Table A4.3) than in Germany (where most of it arises from a discontinuity in 1982–3). In France, there was a slight decrease in this wage ratio during the 1980s, but it too might have increased slightly if controlled for changes in age structure (Cases and Lollivier 1989, table 2). Moreover, the virtual constancy of the high‐to‐low white‐collar wage ratio in France during the 1980s contrasts sharply with its previous steep decline, which started in the late 1960s. In Germany there was also a decline from the early 1960s until the mid‐1970s, and a similar but slighter decline in the USA during most of the 1970s. In the UK, there was a marked dip and recovery during 1974–8, prior to the rise in the 1980s.

(p.442)

Appendix A4 Occupational Relativities

Fig. A4.4 Relative Wages Of Higher and Lower White‐Collar Workers.

Notes and sources: See notes to Fig. A4.1 and Sect. A4.5. To improve readability, 0.5 has been subtracted from the ratio for France, 0.1 from the ratio for Germany, and 0.06 from the ratio for the USA, and 0.1 has been added to the ratio for the UK.

Notes and sources: See notes to Fig. A4.1 and Sect. A4.5. To improve readability, 0.26 has been added to the Canadian ratio, and 0.25 subtracted from the Sweden 1 ratio.

(p.443) Fig. A4.4(b) includes five series covering four other countries. In Austria and Sweden, there appears to have been a modest widening of wage differentials within the white‐collar group during the 1980s, which in the case of Sweden was preceded by a narrowing in the 1970s and little change during the 1960s. In Canada, there was also a decline in the 1970s and a rise in the 1980s (though with a drop in 1986–8, as in the other Canadian series). In Norway, there was a sharp fall in the ratio of higher‐to‐lower white‐collar wages from the late 1960s to the mid‐1970s, but no clear subsequent trend. Age‐controlled calculations for Norway in the 1980s reveal that, as in France, the increasing youthfulness of higher‐level white‐collar groups imparts a downward bias to the all‐age series shown in the graph. In Italy, the higher‐to‐lower white‐collar wage ratio fell steeply between 1974 and 1983, but rose slightly in 1983–6 (OECD‐EO 1987, chart 3.3).

In summary, the behaviour of the more precisely defined occupational wage differentials examined in this section is quite consistent with that of the coarser white‐collar‐to‐blue‐collar ratios looked at earlier. But there are variations among countries in magnitude and in some cases in direction. The UK and the USA apparently experienced the most substantial widening of wage differentials, within as well as between the white‐ and blue‐collar groups. In most of the other countries covered by the data, intra‐white‐collar wage differentials widened after the late 1970s or early 1980s, but by less than in the UK and the USA. However, movements in intra‐blue‐collar differentials were more varied, with no evidence of substantial widening in the 1980s outside the USA and UK, and if anything some narrowing in France and Germany. Even in the UK, the gains of skilled manual workers were largely a recovery of ground lost during the early 1970s, though in Italy similar losses during the 1970s were only partially recouped later.

That many occupational wage relativities in many Northern countries had started to widen or ceased to narrow in the late 1970s and early 1980s was first noted and documented in OECD‐EO (1987, ch. 3). That these ‘new trends’ (ibid. 93) continued into the mid‐1980s was later confirmed for a number of countries by one of the authors of the OECD study (Marsden 1989b). The widening of occupational skill differentials has also been noted in studies of individual countries, particularly of the USA and the UK, where the trend seems to have been most pronounced.

In the USA, Freeman (1987: 16) was apparently the first person to observe that ‘in the 1980s, white‐collar, especially professional, workers have enjoyed much greater wage increases than blue‐collar workers, especially the lower‐skilled (laborers)’, and that this represented ‘a sharp break from historic patterns’. Mishel and Simon (1988, fig. 6) record that the widening gap between white‐collar and blue‐collar pay applied to both male and female workers. Ryscavage and Henle (1990: 12–13) not only confirm that wage gaps among broad occupational categories have widened, but also examine trends in pay for different skill categories within selected white‐collar occupations. In nine out of ten occupations, pay increases were proportionally larger in the higher‐skilled categories, both in the 1980s and in the 1970s, but in only two out of seven occupations in the 1960s.

For the UK, Adams, Maybury, and Smith (1988, charts 1–4) and Katz and Loveman (1990, fig. 7) compare changes in pay with initial levels of pay in more than a hundred detailed occupational categories, for males and females separately. During 1973–9, the association was generally negative, indicating an improvement (p.444) in the relative position of lower‐paid occupations, but during 1979–88 it became strongly positive. Moll (1991, sect. 2.3) confirms that occupational wage differentials in the UK widened throughout the 1980s for both males and females.

A4.3 Relative Unemployment and Vacancy Rates

Data on occupational unemployment and vacancy rates are subject to more than the usual number of problems of interpretation. The most obvious is that unemployed people literally have no occupation. Where they are allocated among occupational categories, this is done either on the basis of their last job or in accordance with their wishes (or an administrative judgement) concerning their next job. Such allocations are unsatisfactory in the present context, partly because higher‐level occupational skills are often specific. For example, a craftsman or manager who has lost his job in a declining industry may be skilled in retrospect and in intention, but in reality have only unskilled job opportunities.10 Occupational vacancy rates do not suffer from this problem of principle, but have several more practical limitations (Walsh 1982).

Jackman, Layard, and Savouri (1991, table 2.4) present two sorts of series of relative unemployment rates by occupation for several countries during the 1970s and 1980s. One refers simply to the ratio of the unemployment rate among blue‐collar workers to that among white‐collar workers. The other sort of series refers to the dispersion of unemployment rates among occupations. These (relative) dispersion indices are calculated with data on a few broad occupational categories (between five and eight categories, depending on the country). The authors interpret these series as showing remarkable stability of relative unemployment rates (ibid. 46). However, regressions against time (whose results are shown in Table A4.2) suggest a different conclusion: both series appear to be rising in all countries but Spain, which is not included in the North in this book.

In the left‐hand half of Table A4.2, the positive coefficients are highly significant in three out of six countries, though barely significant in two of them. These differences in significance levels mainly reflect differences in the size of the coefficients: in Germany and the UK, for example, the ratio of blue‐collar to white‐collar unemployment rates rose over the period by 40 per cent or more, but in Canada by only about 5 per cent. It should be recalled, however, that the white‐collar and blue‐collar groups are both mixtures of skilled and unskilled workers: the true rise in the ratio of the unskilled to the skilled unemployment rate was thus probably greater than the rise in the blue‐collar to white‐collar ratio suggests.

In most countries, the significance levels of the trend coefficients are lower in the right‐hand half of Table A4.2 than in the left‐hand half. However, this could well be because the occupational classifications are not closely related to skill levels. A similar problem of interpretation arises with respect to an earlier study (Jackman and Roper 1987, table 7) which uses occupational unemployment and (p.445)

Table A4.2. Trends in Relative Occupational Unemployment Rates, 1973–1987

Ratio of blue‐ to white‐collar unemployment rate

Dispersion of occupation‐specific unemployment rates

Period

Estimated trend coefficient

Period

Estimated trend coefficient

Australia

1977–86

0.036(*)

1977–85

1.002**

Canada

1975–87

0.011(*)

1975–87

0.093

Germany

1976–85

0.069****

1976–85

0.720**

Spain

1976–87

−0.017*

1977–86

−0.473

Sweden

1973–87

0.022*

1973–84

0.624****

UK

1974–85

0.066****

1974–85

0.268

USA

1973–87

0.046****

1973–82

1.007**

Notes and sources: Based on time series of relative unemployment rates from Jackman, Layard, and Savouri (1991, table 2.4). See present text for a fuller description of the series.

For each country, each of the two relative unemployment rate series was regressed against time (using OLS) over the period shown, which in each case is the longest for which consistent series are available in the source.

The estimated trend coefficients shown above are the coefficients on the independent (time) variables in these regressions. The significance levels of the estimated coefficients (on the basis of t‐tests) are indicated as follows: **** = 1% (two‐tailed test); *** = 2% (two‐tailed test); ** = 5% (two‐tailed test); * = 10% (two‐tailed test); (*) = 10% (one‐tailed test).

vacancy data for eight countries to calculate skill‐mismatch indices (discussed further in Section 8.2). These indices reveal no general tendency for occupational differentials to widen, at least up to 1982 or 1983, when the series end. But again, the occupational classifications involved are not specifically related to skill level, and in some countries are quite detailed (up to forty occupations), which probably exacerbates this problem.11

Another international comparative study of occupational unemployment and vacancy rates, in eight developed countries between the early 1970s and the mid‐1980s, is OECD‐EO (1987, ch. 3, sect. E), some of whose results are further analysed in Marsden (1989a). It notes that relative occupational unemployment rates are cyclical.12 This makes it harder to detect trends, especially because there was a severe recession in 1980–2 and the OECD series stop in the middle of the decade. An additional problem is that some of the series refer to absolute numbers unemployed, rather than to rates of unemployment, and thus understate the (p.446) relative improvement, or exaggerate the relative deterioration, in the position of skilled occupations, whose share of the labour force is rising.

The ratio of professional and managerial to clerical unemployment shows no clear trend in most of the countries, although it declines in France and rises until 1977 in Italy (absolute numbers). The trends in the ratio of white‐collar to blue‐collar unemployment rates are reasonably consistent with the Jackman, Layard, and Savouri (JLS) series described above: there are declines in Australia, France, and the USA, and approximate constancy in Canada and Sweden. The more dubious series based on absolute numbers of unemployed white‐collar and blue‐collar workers give a different impression from the JLS series, though in all cases the discrepancy is in the anticipated direction. Thus the OECD series for Germany is more or less constant, and for the UK rises, whereas the JLS series for both countries imply a steep fall. There is also a rise in the OECD absolute numbers series for Italy, where the data refer only to managers and labourers.

The ratio of skilled to unskilled manual unemployment rates shows little trend in Canada and the USA. In the other four countries in the OECD study with skilled and unskilled manual series, the ratio is based on absolute numbers: it falls in Italy, but rises steeply in France, Germany, and the UK (implying that skilled workers have become relatively more unemployed). These apparent rises are again open to doubt, since they do not allow for the rising ratio of skilled to less‐skilled manual employment, but they might be genuine. In all three of these countries (unlike Canada, Italy, and the USA), the absolute level of blue‐collar employment in manufacturing declined.13 Manual craftsmen with industry‐specific skills in contracting sectors were particularly hard‐hit: reluctant to accept lower‐paid work, and too old to retrain, they had even fewer alternative job opportunities than their less‐skilled colleagues.

This problem of reabsorbing skilled workers displaced from contracting sectors may explain why Micklewright (1984) found in the UK that the rate of unemployment among unskilled manual workers, though high, rose less than in more skilled occupations during 1972–80. During 1982–8, by contrast, the unemployment rates of male unskilled and semi‐skilled manual workers declined less than the average rate (Moll 1991). Skilled manual workers improved their position relative to other manual groups, though not so much as managers, intermediate non‐manual workers, and (especially) professionals. In France, too, unemployment increased more slowly during 1975–84 for professional than for male manual workers (Malinvaud 1987), and ‘the unemployment rate for the unskilled was nearly double the overall rate in 1989 (compared to a gap of only about one‐third in the late 1970s)’.14

The OECD study has data for relative occupational vacancies for only three countries (France, Germany, and the UK). In all cases, moreover, the data refer to absolute numbers, not vacancy rates, which imparts an upward bias to the trend of the skilled‐to‐unskilled ratio because of the rising share of skilled (p.447) employment. In all three countries, the ratio of non‐manual to manual vacancies has an upward trend. In France and Germany, the ratio of professional and managerial to clerical vacancies is also rising, while in the UK it has no clear trend. The ratio of skilled to unskilled manual vacancies is rather volatile, but it appears to have an upward trend in France and the UK, and a downward trend in Germany. This last result is the only one of the nine to imply an improvement in the relative position of less‐skilled workers, although in all cases, as mentioned above, there is a bias in the other direction.

However, there is another study of Germany which does not suffer from this bias (Franz 1991, table 3.3). Defining unskilled workers as those with less than a complete vocational education, it finds that between 1976 and 1987 their share of all vacancies declined faster than their share of total employment, indicating a deterioration in their position relative to skilled workers. In The Netherlands, too, vacancy rates for jobs requiring higher education rose much more in the 1980s than for jobs requiring lower qualifications (van Ours and Ridder 1989, fig. 3.3). This tendency was observed within two separate occupational groups (clerical or business, and technical), whose overall vacancy rates moved in parallel (ibid., fig. 3.1). It suggests that occupational classifications may conceal more than they reveal about shortages and surpluses of labour of differing skill levels.

For the UK, there are also data on the proportion of manufacturing firms which expect their output to be constrained over the next few months by shortages of skilled and unskilled labour. Bean and Pissarides (1991, fig. 7.2) infer from a plot of the annual series that there was only a slight upward shift in the ratio of skilled to unskilled labour shortages in the second half of the 1980s. By contrast, Layard, Nickell, and Jackman (1991, table 19) calculate period averages, and discover a much stronger and longer‐term trend: the skilled‐unskilled shortage ratio rose from 2.7 in 1960–73 to 3.8 in 1974–80 and 6.6 in 1981–9.15

In summary, the data on unemployment and vacancy rates by occupation tend to confirm that the relative position of less‐skilled workers deteriorated in the late 1970s and 1980s. Several series give a different impression, either of no clear trend, or of a contrary trend, but in most of these cases the accuracy or relevance of the underlying data is open to doubt. It is also hard to make inferences from the occupational data about the magnitude of the relative deterioration in the position of less‐skilled workers. This is partly because there are important differences among skill groups within broad occupations, partly because some workers classed as skilled may no longer have marketable skills, and partly because the data do not capture induced exit from (or entry into) the labour force.

A4.4 Demand, Supply, and Employment Trends

With education and age measures of skill, simultaneous changes in the proportions of the labour force in different education and age groups can reveal whether changes in relative wages are caused predominantly by shifts in relative demand (p.448) or by shifts in relative supply (Section A3.3). With the occupational measure of skill, this sort of analysis is not so straightforward. One reason for this is again the difficulty of allocating people who are not working, but are actual or potential members of the labour force, among occupational groups. For at the present time such people have no occupation, and prospectively most of them could work in a wide range of occupations. Moreover, many of those who currently are working in particular occupations form part of the labour supply to other occupations, in the sense that they could and would move if relative wages and job opportunities changed. So although the labour force obviously can be divided exhaustively among a set of mutually exclusive education and age groups, this is in principle not possible for occupational categories.

An alternative approach is to compare relative wage changes instead with changes in relative employment levels in different occupations, which do not suffer from this problem of principle, although in practice comparable wage and employment data are not often available. For movements of relative wages and employment in the same direction clearly tend to imply predominance of demand shifts—and in opposite directions, supply shifts. However, such inferences can be made only when markets clear, or (if they do not clear) when movements in excess demand and excess supply indicators are consistent with the pattern of change in relative scarcities implied by relative wage movements. Thus although there are few actual contradictions of the wage data, the mixed signals from trends in occupational unemployment and vacancy statistics require some caution in interpreting the evidence.16

The share of more skilled occupations in total employment has generally been rising for a long time throughout the North. This trend was documented in an earlier appendix (Table A2.2) for the period between the early 1960s and the mid‐1980s for professional and technical relative to other workers, and for white‐collar relative to blue‐collar workers.17 Within the blue‐collar category, there is less information on the relative employment shares of skilled and less‐skilled workers, and some ambiguity about the (often large) ‘semi‐skilled’ category. But most of the available evidence suggests a long‐term trend in the same direction.18 For example, in the USA, the share of craftsmen among male blue‐collar workers rose from about one‐third in 1900 to just under one‐half in 1982, and the share of (p.449) labourers declined from two‐fifths to under one‐fifth, though with little change between 1960 and 1982. The share of semi‐skilled ‘operatives’ rose until 1960, but then declined.19

What happened during the 1980s is a matter of some controversy, and hard to assess accurately because of changes in occupational classifications and long intervals between censuses. It is agreed by all that the share of white‐collar jobs, and especially of professional and technical jobs, continued to expand. But it has also been argued that the share of middle‐level‐skill jobs declined, and thus that there was a rise in the share of low‐skilled as well as high‐skilled jobs.20 This ‘hollowing‐out’ thesis has been advanced (and disputed) mainly in North America, where the low‐skilled jobs in question are usually said to be in fast‐food restaurants and similar services. However, some data for the UK also suggest a fall in the employment share of middle‐paying occupations, especially among male manual workers in manufacturing, the group whose relative decline has been emphasised also in the US debate (Adams, Maybury, and Smith 1988: 81–2).

The debate over hollowing‐out in the lower reaches of the wage hierarchy will be considered further below (and in Appendix A5). But the evidence on the causes of change in the pay of higher‐level white‐collar occupations, relative to lower‐level white‐collar and to blue‐collar occupations, is fairly clear‐cut. The coexistence during the 1980s of rising relative wages and rising relative employment among professional, technical, and managerial workers strongly suggests a shift in relative demand towards these more skilled groups. Likewise, the decline in their relative pay in most countries during the 1970s, when their share of total employment was also rising, strongly suggests that the predominant influence then was an increase in the relative supply of these more skilled workers. In neither decade do changes in relative unemployment and vacancies contradict the wage movements. And in both decades, the pattern is consistent with (and closely related to) that for college graduates described in Appendix A3.

The change in the fortunes of the higher‐level occupations in the 1980s appears more striking when it is recognised that the 1970s combination of declining relative pay and rising relative employment represented the continuation of a much longer‐term trend. This trend was neither uniform across occupations nor smooth across decades. But for most professions, and probably also for managers, it had apparently been the usual direction since at least the beginning of the twentieth century, and as in the 1970s was probably caused largely by the relative expansion of upper‐secondary, technical, and higher education. (The introduction of (p.450) free and compulsory lower‐secondary schooling had also contributed to a long‐term decline in the pay of clerical relative to blue‐collar workers.) Among manual workers, too, the long‐term increase in the proportion of skilled jobs had generally been associated with a secular narrowing of skill differentials in wages. (See for example the studies surveyed in Wood 1978: 181–202; Williamson and Lindert 1980; and Saunders and Marsden 1981.)

As explained earlier, movements in the relative pay of higher‐level white‐collar workers were the main ingredient of movements during the 1970s and 1980s in the average white‐collar to blue‐collar wage ratio. So the assessment above that a supply shift narrowed higher‐level occupational skill differentials in the 1970s, while the widening in the 1980s was caused by a demand shift, must apply also to the coarser wage relativity between white‐collar and blue‐collar workers. The white‐collar share of employment increased during both decades, which is consistent with this assessment. Moreover, trends in the relative unemployment and vacancy rates of white‐collar and blue‐collar workers, though somewhat mixed, do not generally contradict this interpretation of the evidence.

However, in some countries there are clear indications, both for white‐collar workers in general and for higher‐level groups, that institutional forces also had an effect on relative wage movements (Marsden 1989b: 3.5–10). In the United Kingdom during the 1970s, when incomes policies discriminated against higher‐paid workers, Figs. A4.1 and A4.4(a) show sharp dips in the relative pay of white‐collar and professional workers. In Italy, the reversal of the downward trend in the relative pay of these two groups started when the official system of wage indexation was amended in 1983. In France, minimum wage (SMIC) policy, and in Sweden union wage bargaining policies, may have contributed to the observed narrowing of these wage differentials during the 1970s and much of the 1980s. And in other countries it seems likely that the tendency for these differentials to widen during the 1980s was inhibited, albeit to varying degrees, by some combination of union action by lower‐paid white‐collar and blue‐collar workers, minimum wage legislation, and social security income floors.

It is harder to assess the causes of the observed behaviour of wage relativities between skilled and unskilled blue‐collar workers during the 1970s and 1980s, even in particular countries, and impossible to generalise about the North as a whole. For these intra‐blue‐collar wage differentials moved in different ways in different countries (Fig. A4.3). The evidence on changes in relative unemployment and vacancies for these occupational groups is likewise mixed. And there is less information, and more dispute (connected with the hollowing‐out argument), about trends in the relative employment of more and less skilled manual workers. Some conclusions may none the less be extracted from the limited data available.

Table A4.3 shows employment and wage changes in selected occupations in the USA during the 1980s. The above‐average increases in employment and wages for managerial, professional, and technical workers are apparent for both males and females. The table also confirms that, among blue‐collar males, the wages of craftsmen went up faster than those of machine operators, who in turn did better than labourers. (The corresponding ranking for females is less meaningful, since there are so few in the highest and lowest categories.) However, the relative employment levels of craftsmen and machine operators did not change. Moreover, employment among labourers increased relative to these two other (p.451)

Table A4.3. Employment and Wages in Selected Occupations in the USA in the 1980s (Full‐Time Wage and Salary Workers)

Wage level 1990 (ratio of handlers etc.)

Employment share 1990 (%)

Employment increase (1990 level/1984 level)

Money wage increase (1990 level/1984 level)

Males

Executive, administrative, managerial

2.45

12.9

1.08

1.32

Professional specialty

2.37

12.1

1.10

1.33

Technicians and related support

1.85

3.5

1.17

1.28

Administrative support, including clerical

1.44

6.6

1.08

1.17

Precision production, craft, and repair

1.57

20.6

1.00

1.24

Machine operators, assemblers, and inspectors

1.25

9.2

1.00

1.20

Handlers, equipment cleaners, helpers, laborers

1.00

6.0

1.18

1.16

Service except household and protective

0.91

6.1

1.26

1.29

Other occupations*

n.a.

23.0

1.07

n.a.

Total

n.a.

100.0

1.07

n.a.

Females

Executive, administrative, managerial

1.92

13.3

1.46

1.34

Professional specialty

2.11

16.5

1.27

1.38

Technicians and related support

1.60

3.8

1.20

1.28

Administrative support, including clerical

1.29

30.9

1.07

1.28

Precision production, craft, and repair

1.21

2.4

1.02

1.20

Machine operators, assemblers, and inspectors

1.01

7.8

0.93

1.20

Handlers, equipment cleaners, helpers, laborers

1.00

1.9

1.23

1.23

Service except household and protective

0.90

11.1

1.21

1.27

Other occupations*

n.a.

12.2

1.18

n.a.

Total

n.a.

100.0

1.16

n.a.

(*) Sales, private household and protective service, transport, farming.

Source: US Bureau of Labor Statistics, Employment and Earnings, Jan. issues, table A‐75. Data refer to fourth quarter: apparently not published in comparable form prior to 1984 (occupational classification changed in 1982). ‘Wages’ are median weekly earnings.

blue‐collar groups, which seems inconsistent with a shift of relative demand against the unskilled. And for (male and female) service workers, the lowest‐paid group, both wages and employment increased relative to other manual occupations.

The table clearly gives some support to the hollowing‐out hypothesis. Among both males and females, employment in the two lowest‐paid groups (labourers and service workers) expanded faster than for the three groups immediately above them (clerical workers, craftsmen, and machine operators). But its implications regarding the causes of the widening of skill differentials in wages among blue‐collar workers are less clear, because the occupational classification is a mixture of skill‐related and sectoral characteristics. For both males and females, the relative employment decline has been most pronounced in the two occupational categories most closely associated with manufacturing (craftsmen and machine operators).

In the UK, the other country where wage differentials among manual workers clearly widened during the 1980s, the occupational classification does not provide a general categorisation of blue‐collar workers by skill (the wage differential in Fig. A4.3 refers to two narrow occupations). As in the USA, the share of employment (p.452) in occupations associated with manufacturing (groups XII to XV) declined during the 1980s, from 35 per cent to 30 per cent for males and from 18 per cent to 13 per cent for females.21 Unlike the USA, however, the share of total employment in the low‐paid catering, cleaning, and personal service category increased only slightly for males and declined for females. This might reflect the relatively higher reservation wage of less‐skilled labour in the UK, because of the general income floor provided by the social security system (which has no counterpart in the USA). But the differences between the occupational classifications preclude a proper comparison of the degree of hollowing‐out in the two countries.

Layard and Nickell (1986: 40–1) combine data from different sources to show that the relative demand for less‐skilled manual workers declined sharply in the UK during 1979–85. They infer from Micklewright (1984) that the unemployment rate among semi‐skilled and unskilled manual workers is roughly double that among skilled manual workers, and that these two rates have risen over time by roughly the same proportion. This implies that employment of less‐skilled manual workers has fallen by proportionally twice as much as employment of skilled manual workers. Since the relative wage of less‐skilled workers has also fallen, as evidenced by the increased dispersion of manual wages, relative demand must have shifted against them. Assuming an elasticity of substitution between skilled and less‐skilled workers of 2.5 (on the basis of another study), Layard and Nickell estimate the size of the shift during this period at one‐third, which they rightly describe as ‘huge’. If, as argued in Section 6.1.2, the elasticity were about unity, the implied shift would be smaller (about one‐fifth), but still substantial. Moreover, the unemployment rate of less‐skilled workers actually deteriorated relative to skilled manual workers during the 1980s (Moll 1991), implying an even faster relative decline in their employment.

This decline in the employment of less‐skilled relative to skilled manual workers in the UK is in contrast to the USA, where, among males, employment of craftsmen did not change relative to machine operators, and declined relative to both labourers and service workers. But the limited data available for other Northern countries generally show the same trend as in the UK. In France the ratio of skilled to semi‐ and unskilled male manual employment in the private and semi‐public sectors rose from 1.22 in 1970 to 1.66 in 1980 and 1.73 in 1985, with a similar trend for females (Marsden 1989a, table 6.3). In Germany the share of skilled workers in male manual employment in industry increased steadily from 41 per cent in 1970 to 48 per cent in 1985, but for females remained constant at about 1 per cent (ibid.). In Sweden the ratio of skilled to non‐skilled full‐time manual employees rose from 0.65 in 1975 to 0.84 in 1984 (Marsden 1989b, table 2.9). In Austria this ratio also went up between 1983 and 1989 for females, but declined for males.22 In Denmark the male manual skilled–unskilled ratio in manufacturing and construction appears to have been constant from 1975 to 1988, perhaps because of the sampling procedure used.23

(p.453) In these other countries, however, there was no general or strong tendency for wage differentials among manual workers to widen. In principle, this association between rising relative employment and unchanging relative wages of skilled manual workers might reflect a declining relative supply of less‐skilled workers. But this interpretation is not at all plausible, given the high levels of unemployment among less‐skilled workers in most of these countries. For even though the evidence on changes in relative unemployment and vacancy rates for skilled and less‐skilled manual workers is mixed (as noted earlier), the undisputedly much higher unemployment rates among less‐skilled than among skilled manual workers make it hard to believe that increasing scarcity of less‐skilled workers could be behind the decline in their employment with maintenance of their relative wage. A more plausible explanation is that the lack of change in the relative wage reflects institutional pressures, while the relative reduction in less‐skilled employment reflects an adverse movement of relative demand.

To sum up, the evidence on trends in relative wages and employment by level of skill among blue‐collar workers is scanty, open to various doubts, and does not all point in the same direction. But most of it seems consistent (like the evidence on other occupational categories) with the hypothesis of an increase in the relative demand for skilled workers, in conjunction with differing degrees of relative wage flexibility. In the USA, the shift in demand appears to have emerged mainly as a widening of wage differentials, perhaps facilitated by the relative decline of the federal minimum wage after 1981, rather than as a change in relative employment.24 In most European countries, the opposite seems to have happened, with rather rigid wage differentials, and a decline in less‐skilled employment. In the UK, powerful institutional constraints on relative wages were deliberately weakened during the 1980s, but by no means eliminated, so that among blue‐collar workers there was both a widening of skill differentials and a decline in less‐skilled employment. What happened in Japan is not known.

A4.5 Annexe: Definitions and Sources of Wage Data

This section contains detailed information, organised by country, on the data used in compiling the figures in earlier sections (and Table A4.1). In many cases, the figures are updated versions of those in OECD‐EO (1987, ch. 3), with no change in definitions or sources. More detail is provided below where the coverage, definitions, or sources differ from those in the OECD study. Table numbers are cited for only one recent (usually 1990) issue of annual national statistical year‐books, although some of the data may have been drawn from corresponding tables in earlier issues. The labels (a), (b), and (c) refer to the same occupational groupings in all countries, though data on all three are not available in all countries.

Austria

The data are from various issues of the annual Statistiches Handbuch für die Republik Österreich.

(p.454)

  1. (a) White‐collar and blue‐collar workers. An update of the OECD series, which refer to the monthly earnings of male Angestellte and Arbeiter in industry. (Table 9.06 of the 1990 Handbuch.)

  2. (c) Higher and lower white‐collar workers. Median standard net personal income of male Angestellte mit hochqualifizierter Tätigkeit and Angestellte mit gelernter Tätigkeit in all sectors, in 1983 and 1989 (interpolated). (Table 9.24 of the 1990 Handbuch.)

Belgium

The data are from various issues of the annual Statistiques Sociales.

  1. (a) White‐collar and blue‐collar workers. An update of the OECD series, which refer to indices of the monthly salaries of male employés and the hourly earnings of male ouvriers in industry. (1990 issue no. 3, p. 70.)

Canada

The data are from the Survey of Consumer Finances, and are drawn from a special tabulation provided by the Analytical Studies Branch of Statistics Canada. They refer to full‐time full‐year males in all sectors, and to annual labour income, which includes self‐employment income as well as wage and salary income.

  1. (a) White‐collar and blue‐collar workers. Blue‐collar workers are non‐farm manual, and white‐collar workers a fixed‐weighted average of professional and technical and clerical.

  2. (c) Higher and lower white‐collar workers. Professional and technical workers, and clerical workers.

Denmark

The data are from various issues of the annual Statistisk Årbog.

  1. (a) White‐collar and blue‐collar workers. Monthly earnings of male salary‐earners in all sectors, and hourly earnings (including overtime) of male manual workers in manufacturing and construction. (Tables 201 and 203 of the 1990 Årbog.) Extended backwards in 1970–5 using OECD series, which is based on salary and wage indices for both sexes in manufacturing.

  2. (b) Skilled and unskilled blue‐collar workers. Hourly earnings (including overtime) of skilled (mainly male) and unskilled male wage‐earners in manufacturing and construction. (Table 201 of the 1990 Årbog.)

Finland

The data are from various issues of the annual Statistisk Årsbok.

  1. (a) White‐collar and blue‐collar workers. An update of the OECD series, which refer to indices of the monthly earnings of salaried employees and the hourly earnings of wage earners of both sexes in all sectors. (Table 350 of the 1990 Årsbok.)

(p.455) France

The data are from various issues of the Annuaire Statistique de la France, Economie et Statistique, Marsden (1989b, table 2.7), Lhéritier (1990), and Girard and Lhéritier (1991).

  1. (a) White‐collar and blue‐collar workers. A modification (males rather than both sexes, except in 1975–80) and update of the OECD series. Annual earnings of full‐time workers in private and semi‐public sectors. White‐collar workers are a fixed‐weighted average of cadres supérieures, cadres moyens, and employés; blue‐collar workers are ouvriers. After 1985, the series is not strictly comparable: the occupational classification changes, and the data apparently refer only to the private sector.

  2. (b) Skilled and unskilled blue‐collar workers. Ratio of annual earnings of male ouvriers qualifiés to manoeuvres in 1980–5 in private and semi‐public sectors (table C.03–2 of 1986 Annuaire), extrapolated back to 1973 using hourly wage indices for level 5 and level 1 manual workers (both sexes 1977–80, males 1973–7), and then back to 1964 using hourly wage indices for male ouvriers qualifiés and level 1 manoeuvres. Extrapolated forward to 1990 using data on the annual earnings of male ouvriers qualifiés and non‐qualifiés.

  3. (c) Higher and lower white‐collar workers. As for (a), using the ratio of cadres supérieures to employés. Extrapolation after 1985, when coverage of cadres alters.

Germany

The data are from various issues of the annual Statistiches Jahrbuch.

  1. (a) White‐collar and blue‐collar workers. An update of the OECD series, which refer to the monthly earnings of male Angestellte and Arbeiter in industry. The updating used indices of monthly earnings of Angestellte and weekly earnings of Arbeiter. (Tables 22.2 and 22.6 of the 1990 Jahrbuch.)

  2. (b) Skilled and unskilled blue‐collar workers. From Marsden (1989b, table 2.8, males), interpolated and updated using data on the weekly earnings of Leistungsgruppe 1 and 3 of male Arbeiter in industry. (Table 22.3 of the 1990 Jahrbuch.)

  3. (c) Higher and lower white‐collar workers. An update of the OECD series, which refer to the monthly earnings of male Angestellte in industry, trade and finance. Ratio of Leistungsgruppe II to V. Extrapolated to 1987 from Marsden (1989b, table 2.8), and to 1989 using average wage increases of kaufmannische and technische Angestellte. (Table 22.7 of 1990 Jahrbuch.)

Italy

  1. (a) White‐collar and blue‐collar workers. An update of the OECD series, which refer to the average annual earnings of impiegati and operai of unspecified sex in manufacturing. This series was converted into an index and extrapolated to cover 1971–4 and 1986–9 using indices of earnings for unspecified pay‐periods for workers of unspecified sex in industry. (Table 18.1 of the 1990 Annuario Statistico Italiano.)

  2. (p.456)
  3. (b) Skilled and unskilled blue‐collar workers. This is the OECD series, which refer to the typical annual pay of Livello 5 and Livello 2 manual workers of unspecified sex in manufacturing.

Japan

  1. (a) White‐collar and blue‐collar workers. This is the OECD series, which refers to the monthly earnings of male non‐production (or salaried) and production workers in manufacturing.

Norway

The data are from various issues of the annual Statistisk Årbok.

  1. (a) White‐collar and blue‐collar workers. An update of the OECD series, which refer to (white‐collar) a fixed‐weighted average of the monthly earnings of male technical, supervisory, office, and warehouse employees, mainly in manufacturing, and (blue‐collar) the hourly earnings of adult male manual workers in manufacturing. (Tables 189 and 194 of the 1990 Årbok.) The 1987–9 values of the index were adjusted for a reduction in weekly working hours (with full compensation) from 40 to 37.5 in January 1987. (OECD Economic Survey of Norway 1987–8: 75.) Had this adjustment not been made, the index would have dropped sharply (and misleadingly) between 1986 and 1987.

  2. (c) Higher and lower white‐collar workers. Also an update of the OECD series, using the ratio of male managers (chief engineers and those in management positions) to simple routine workers, mainly in manufacturing. (Table 194 of the 1990 Årbok.)

Sweden

  1. (a) White‐collar and blue‐collar workers. For 1960–80, the OECD series, which refer to salaries (presumed monthly) of full‐time male technical and office staff, and hourly earnings of male workers in manufacturing. Extrapolated 1980–8 using data on the annual earnings of full‐time males in the public and private sectors; white‐collar workers are a fixed‐weighted average of högre tjänstemän and tjänstemän pȧ mellannivȧ, and blue‐collar workers are arbetare och lägre tjänstemän. (Data from table 25 of the Income Distribution Survey 1988.)

  2. (c) Higher and lower white‐collar workers. Series 1 (1960–84) is the OECD series, using the ratio of male managerial to clerical workers in manufacturing. Series 2 (1975–88, interpolated) is the ratio of the annual earnings of male högre tjänstemän to those of tjänstemän pȧ mellannivȧ in the public and private sectors. (Table 25 of the Income Distribution Survey 1988.)

Switzerland

The data are from various issues of the annual Statistiches Jahrbuch.

  1. (a) White‐collar and blue‐collar workers. An update of the OECD series, which (p.457) refer to indices of the monthly salaries of Angestellte and the hourly wages of Arbeiter of both sexes in industry. (Table 3.16 of the 1991 Jahrbuch.)

The Netherlands

  1. (a) White‐collar and blue‐collar workers. An update of the OECD series, which refer to the weekly earnings of male employés and arbeiders in industry. The update used indices for non‐manual and manual workers of both sexes. (Statistical Yearbook of the Netherlands 1988, sect. V, table 3.)

United Kingdom

The data are from various issues of the annual New Earnings Survey, and refer to the median weekly earnings of full‐time adult males whose pay was not affected by absence, in all sectors.

  1. (a) White‐collar and blue‐collar workers. Non‐manual and manual workers. Includes overtime pay. (From the summary volume of the NES.)

  2. (b) Skilled and unskilled blue‐collar workers. Maintenance fitters (non‐electrical) from group XIV, and general labourers (including engineering and shipbuilding) from group XVIII. Excludes overtime pay after 1971. (From table 98 of Vol. D of the 1990 NES.)

  3. (c) Higher and lower white‐collar workers. Professional and related supporting management (group II) and clerical and related (group VII). Excludes overtime pay. (From table 98 of Vol. D of the 1990 NES.)

United States of America

  1. (a) White‐collar and blue‐collar workers. For 1967–80, the OECD series, which refer to the annual earnings of full‐time males in all sectors. Extrapolated to 1990 using private industry employment cost index data from Freeman (1987, table 1.8), Ryscavage and Henle (1990, table 11), and Monthly Labor Review (Dec. 1990, table 22).

  2. (b) Skilled and unskilled blue‐collar workers. For 1970–83, the OECD series, which refer to the median weekly earnings of craft and kindred workers and non‐farm labourers of both sexes in all sectors. Extrapolated forward using data from Current Population Survey (1984–7) and Employment and Earnings (1988–90) on ‘precision production, craft and repair’ and ‘handlers, equipment cleaners, helpers and laborers’. Extrapolated backwards to 1962 using data on the median annual earnings of males from the Statistical Abstract.

  3. (c) Higher and lower white‐collar workers. Professional and clerical workers of both sexes in all sectors, from Freeman (1987, table 1.8), interpolated and extended to 1990 using data from Employment and Earnings for ‘professional specialty’ and ‘administrative support including clerical’. Extrapolated backwards to 1962 as in (b), using data on ‘professional and technical’ and ‘clerical and kindred’. (Freeman bridges the occupational classification change in 1982.)

Notes:

(1) In the USA and Canada there are also estimates of the amounts of General Educational Development and Specific Vocational Preparation required in detailed occupational categories (see e.g. Myles 1987).

(2) Since the OECD study covered occupational differentials in general, this indebtedness applies also to the following sections. David Grubb, one of the authors of the study, kindly supplied much of the underlying data. David Marsden, the other author, generously provided some recent unpublished data of a similar kind (Marsden 1989b). As is explained in Sect. A4.5, it was possible in some cases simply to update the OECD series from the original sources, but in other cases they had to be extended (and sometimes replaced) with data from different sources.

(3) This depends of course on how the indices are constructed, which is rarely specified in the sources on which these graphs are based.

(4) The change in trend in 1983 coincided with the partial de‐indexation of wages (Marsden 1989b: 2.4). However, the 1986–9 data are not comparable with those for 1974–86, and may understate the rise. Like the data for 1971–4, they are based on wage index series whose pay‐period basis is not stated in the source, but may well refer to monthly salaries and hourly wages. The 1974–86 data refer to annual earnings, and thus are not affected by changes in working hours.

(5) Jackman, Layard, and Savouri (1991, table 2.16) present similar data on white‐ and blue‐collar wages for seven European countries, but conclude that in general skill differentials did not widen during the 1980s.

(6) This decline and recovery can be seen also in chart 3.4 of OECD‐EO (1987), which is based on a different (and no longer published) source of data.

(7) See chart 3.4 of OECD‐EO (1987).

(8) Of the three alternative series for Italy cited by Marsden (1989b), one rose in the 1980s (ibid. 2.4), and two did not (ibid. 2.4 and table 2.6).

(9) The median standard income of Facharbeiter as a ratio of that of Hilfsarbeiter in the private sector was 1.15 for males in both 1983 and 1989. For males and females together, this ratio declined from 1.29 in 1981 to 1.23 in 1989. (Calculated from table 9.24 of the 1990 Statistiches Handbuch für die Republik Österreich and similar tables in earlier issues.)

(10) Nor is it meaningful with occupational categories (unlike education and age groups) to calculate employment‐to‐population ratios to overcome the limitations of official unemployment statistics.

(11) The 327–occupation calculation for Germany by Franz (1991: 119–21) is a more extreme illustration of this point. It suggests a decline in skills mismatch, whereas his more reliable two‐category education‐based calculation shows a marked increase in mismatch.

(12) The cyclical pattern has several different causes. One is arithmetic: uniform proportional reductions in employment cause larger proportional increases in unemployment rates in occupations where the unemployment rate was initially lower (see n. 14, Ch. 8). Thus recessions tend to raise relative unemployment rates in more skilled occupations. This effect is compounded within the blue‐collar group by the fact that recessions tend to displace fewer unskilled than skilled manual workers, since the skilled workers are concentrated in two cyclically sensitive sectors—construction and manufacturing. However, white‐collar workers tend to be less affected by recessions than blue‐collar workers.

(13) See fig. 3 of Wood (1991a). This figure does not include Canada, where the absolute level of manufacturing employment increased by more than 50% over the period, with little change in the ratio of white‐ to blue‐collar workers (see Table A2.2), implying a similar proportional increase in the absolute level of blue‐collar employment.

(14) The quotation is from the OECD Economic Survey of France (1990–1: 56). ‘Unskilled’ is not defined.

(15) The 1960–73 and 1981–9 figures given here are weighted averages of two sub‐period figures in the original source. This averaging does not affect the conclusion.

(16) Marsden (1989a) finds only weak evidence of an inverse short‐term association between changes in the relative pay of different occupational groups and changes in their relative unemployment levels or rates.

(17) The last two columns of Table A2.2 show increases in the professional and white‐collar employment ratios of manufacturing in all countries. It can be inferred from the table that these ratios rose also in nontraded sectors, except for the professional ratio in the USA and New Zealand. The table refers to changes in specific sectors, and therefore understates the economy‐wide increases in these two ratios, since it does not capture the effects of the increased share of total employment in nontraded sectors, where both ratios are relatively high.

(18) In the UK, the share of skilled jobs in total manual employment was 42% for males both in 1911 and in 1951, but had risen to 51% in 1970 (Routh 1965, table 1; New Earnings Survey 1970, table 162; classifications only roughly comparable). Among females, this share declined from 30% in 1911 to 20% in 1951, and had risen only to 23% by 1970. The share of skilled workers in total male manual employment in the UK engineering industry rose from 46.7% in 1970 to 49.0% in 1980 (Marsden 1989a, table 6.A.3). In France, population census data show that the share of qualified workers in total manual employment (ouvriers and manoeuvres of both sexes) rose from 38% in 1962 to 55% in 1982. Between 1972 and 1978, the ratio of skilled to total manual employment rose among males in Belgium, France, Germany, and Italy, but fell in The Netherlands (OECD‐EO 1987, table 3.5). For females, it rose in France and The Netherlands, remained constant in Germany, and fell in Belgium and Italy. Other more recent data for various countries are cited later in the text.

(19) Data for 1900–60 refer to the labour force and are from Historical Statistics of the USA (1975, Series D 182–232). Data for 1960–82 refer to employment and are from Statistical Abstract of the USA (1984, table 693). In both cases, manual workers exclude farm and service workers, and operatives include transport workers. The occupational classification differs from that in Table A4.3.

(20) Recent contributions to the debate, with references to earlier studies, include Wolfson (1989), Harrison and Bluestone (1990), and Picot, Myles, and Wannell (1990).

(21) These results are from an earlier version of Moll (1991), and are based on data from the New Earnings Survey.

(22) Source as in Sect. A4.5. The same result obtains whether ‘skilled’ workers are defined to include only Facharbeiter or also Meister und Vorarbeiter.

(23) Source as in Sect. A4.5. The share of skilled workers in the total is between 41% and 43% throughout the period.

(24) On the trend in the minimum wage, see Mishel and Simon (1988: 43) and Marsden (1989b: 3.7).