glyn (1995) - the assessment - unemployment and inequality

OXFORD REVIEW OF ECONOMIC POLICY, VOL. 11, NO. 1

THE ASSESSMENT:UNEMPLOYMENT AND INEQUALITY

ANDREW GLYNWissenschqftszentrum, Berlin, and Corpus Christi College, Oxford*

I. INTRODUCTION

International organizations are currently subject-ing labour-market trends in the advanced countriesto an unprecedented barrage of analytical scrutinyand policy prescription (Commission of the EC,1993;OECD, \994a,b; ILO, 1995). Figure 1, liftedfrom the OECD's massive Jobs Study, illustrateswhy. Unemployment fell only modestly in the ex-pansion of the later 1980s and its subsequent furtherrise could not, once again, be blamed primarily onexternal 'shocks'. Several countries, most notablySweden, which seemed to have found ways ofcoping with slower growth without rising unem-ployment, have finally succumbed. Others, like theUK and New Zealand, which most rigorously ap-plied the free-market principles of the 1980s, havefailed to register marked improvements in labour-market outcomes.

Since unemployment was the subject of a completeissue of this Review (voi. 1, no. 2, 1985), it hasincreasingly come to be seen in the context of anoverall increase in labour-market inequality. Thisincludes, along with rising registered unemploy-ment, withdrawal from the labour force (economic'inactivity') and greater earnings inequality. In thisbroader context, policies have to be examined notonly for their effectiveness in reducing unemploy-ment, but also for the distribution of costs andbenefits. This assessment, and the issue as a whole,reflects such a broadening of the debate.

The second section outlines in very summary formthe main underlying trends in the OECD econo-mies, to give the context in which the rise inunemployment has occurred. Then the main fea-tures of the unemployment position are documented,and placed in relation to other dimensions of la-

1 I would like to thank Brian Bell and Gregory Wurzburg for providing data, Wendy Carlin, Ken Mayhew, and Ronald Schettkatfor very helpful comments, and Andrea Boltho, Paul Gregg, and Bob Rowthorn for earlier collaborations which contributed a greatdeal to this paper.

© OXFORD UNIVERSITY PRESS AND THE OXFORD REVIEW OF ECONOMIC POLICY LIMITED

by guest on January 17, 2011oxrep.oxfordjournals.org

Dow

nloaded from

http://oxrep.oxfordjournals.org/


Figure 1Unemployment in the OECD Area, 1950-93

35 T

30

25 +S| 20

I15

1 0 V

First oil-price shock

Second oil-price shock

cs t vo oot— r- r— f»ON ON ON ON

OooO

oo ooO\ ON

ig oooo ooON ON

year

Afote: Including eastern Germany from 1991 onwards.Source: The OECD Jofo Study.

bour-market inequality. The fourth section presentsalternative views as to the connections betweenunderlying labour-market trends and rising labour-market inequality: is the latter simply the form thatcertain underlying macroeconom ic forces take (suchas a higher non-accelerating inflation rate of unem-ployment (NAIRU)) or is it evidence of structuralproblems which require a broadening of the analy-sis? The conclusion outlines the main lines ofpolicy response.

II. MACROECONOMIC ANDSTRUCTURAL TRENDS

(i) Jobless Growth?

The combination of stubbornly high unemploy-ment and the spread of computer-based technologyhas led to a belief that economic growth no longerrequires additional jobs to the extent characteristicof the 'golden age' of growth in the 1950s and1960s. This is very misleading. Despite the halvingof output growth in the OECD after 1973, numbersat work actually rose faster, up to the recession ofthe early 1990s, than before 1973 (Table 1).

The surprising finding of faster employment growthafter 1973 is true of the EC where employment

problems have been especially severe. Since aver-age hours of work seem, from patchy data, to havebeen declining less rapidly in Europe than before1973, there is no evidence of a slowdown in theamount of work performed. Total labour input inthe EC was roughly constant over the period 1979-90, while it probably declined at around 0.5 percentper year over the period 1960-73.

The relative steadiness of employment growth im-plies that labour productivity growth has declinedas much (or even more in hourly terms) than outputgrowth. It is absolutely wrong, therefore, to blametechnology for 'destroying jobs' at an unprec-edented rate; if this had been the case, labourproductivity would have been rising faster thanbefore.

In the short term, rapid expansions in output stilllead to faster job creation. In the UK for example,the period 1985-8 saw output rising by 4.7 per centper year and employment grew by 2.0 per cent peryear; from 1990 to 1992 output fell at 1.4 per centper year and employment declined by 3 per cent peryear. With labour productivity typically rising inOECD countries in the range 1.5-2.5 per cent peryear in sustained expansions, maintaining a 4 percent per year growth for several years provides jobsat a rate of around 1.5-2.5 per cent per year.


Dow

nloaded from


A. Glyn

Table 1Employment Growth, 1960-93 (Average Annual Percentage Changes)

OECD EC USA Japan

1960-731973-901990-3

1.11.2

-0.1

0.30.5

-1.1

2.01.90.3

1.31.01.1

Source: OECD Historical Statistics, Economic Outlook, Employment Outlook.

Conversely, severe recessions inevitably bring ma-jor job cuts and this has been especially true of therecession of the early 1990s (the most spectacularcase being the 18 per cent fall in Finnish employ-ment from 1990 to 1993). The Jobs Study noted(OECD, 19946, p. 55) thatthe rather weak recoveryin employment at present reflected a feeble reboundin output rather than a failure of employment torespond. The impact of output on jobs applies in themedium-term as well. A cross-section analysis ofthe (trough-to-trough) period 1982-93 found thatfor every 1 per cent per year that a country increasedits output growth (as compared to the previouscycle), there was a nearly 1 per cent boost toemployment growth (Boltho and Glyn, 1994, Table3).

(ii) The Pattern of Jobs

Between 1973 and 1990 the share of OECD indus-trial employment (manufacturing, mining, construc-tion, and utilities) declined from 36 per cent to 30per cent. In the EC the shift out of industry has beenrather faster, starting earliest in the UK where it fellto 29 per cent by 1990 as compared to 47.7 per centin 1960.

Industrial productivity typically grows faster thanthe average forthe whole economy because of weaktechnical change in services (see Baumol et al.,1989). Thus a constant share of industrial employ-ment requires that industrial output also growsfaster. After 1973, however, industrial output grewmore slowly than GDP (by 1.5 per cent per year inthe EC, for example, as compared to 2.3 percent forGDP). More than a switch of consumer demandtowards services was involved. Investment, whichis mainly demand for industrial output, stagnatedup to 1979. Subsequently, net exports of manufac-

tures declined, from 14 per cent to 5 per cent ofvalue added in EC manufacturing between 1979and 1990. Combined with the usual faster growth oflabour productivity in industry the result of the slowgrowth of industrial output was a declining share ofindustrial employment.

The rate of outflow of workers from industry since1973 has been less than from agriculture before1973. So why should loss of industrial jobs besingled out? The explanation is in the nature ofindustrial work. Traditionally in the OECD coun-tries this has been relatively well paid, mainlycarried out by men, working full-time with skillsthat are specific to industrial work and requiringonly basic education as a prerequisite. Major de-clines in industrial employment result in large-scale, geographically concentrated redundancieswhich flood local labour markets with less-edu-cated labour. It is difficult for service employmentto take up the slack. Much of it (public services,retail and wholesale distribution, and personal serv-ices) supplies a local population and is thus spreadrelatively uniformly across the economy. Even thoseproducer services which are more mobile oftenrequire different skills and working schedules fromindustrial work, with the result that service employ-ment may be unable to absorb many of those losingjobs in the industrial sector. By contrast, the out-flow from agriculture before 1973 occurred in thecontext of plentiful job opportunities in the towns.Particularly in the early 1980s, many of the ECcountries experienced a collapse of industrial em-ployment (in the EC it fell by 13 per cent between1980 and 1985) which was immediately reflected inrising unemployment despite the fact that employ-ment in services rose by nearly 8 per cent (a similaramount in absolute terms). Many of the new servicejobs were filled by women entering the labour


Dow

nloaded from


OXFORD REVIEW OF ECONOMIC POLICY, VOL 11, NO. 1

Table 2Sector Per Capita Employment Growth, 1973-90 (Average Annual Percentage Changes)

Agriculture Industry Services Total

USAJapanECUK

Note: Per capita employment growth is growth of employment less growth of population of working age.Source: OECD Historical Statistics.

-1.6-3.7-3.9-1.9

-0.7-0.4-1.6-2.2

1.51.21.31.4

0.70.1

-0.20.0

force, leaving a large pool without work, includingmany of those previously with industrial jobs.

It is often presumed that a less dynamic creation ofservice jobs is what lies behind the relatively slug-gish growth in European employment as comparedto the US 'Jobs Machine'. The best way to assess asector's contribution to providing work is to com-pare employment growth with population of work-ing age. This (Table 2) shows that since 1973services in the EC have been nearly as effective inproviding jobs for the working population as theywere in the USA; in fact in the 1980s their rates ofemployment creation were the same. The differ-ence in employment performance is almost entirelydue to the much faster run-down of employment inagriculture in Europe (which fell by about 3.5 percent of the population of working age in the EC) andthe very rapid decline in industrial employment. Inthe UK, which started with a smaller agriculturalsector than the US, the difference in employmentperformance is entirely a reflection of the loss ofindustrial jobs. Indeed in the 1980s per capitacreation of service jobs was faster in the UK. Giventhe fasterrundown of agriculture in most of Europeand the decline in industrial employment since1973, the expansion of services employment neededto be, but was not, faster than in the USA.

The service sector is such a heterogeneous collec-tion of activities that it is helpful to disaggregate.Table 3 shows the importance of the major servicesectors; finance and business services and the pri-vate part of community, social, and personal serv-ices (CSPS) have represented the major sectors foremployment growth, and here the EC's record since1973 is very close to that of the USA. But stillemployment shares are well below those of the US,

as is true also for wholesale and retail trade. Moreo-ver, the shortfalls are several percentage pointslarger in both trade and CSPS if employment iscompared to the whole population as indicatingpotential need for these services (the ratio of em-ployment to population is nearly 15 per cent higherin the USA). But it should be remembered that the'service economy' in the USA developed over avery protracted period. The share of industrialemployment only began to decline in the early1970s, when agriculture had already shrunk to alow level. In Europe, industrial decline has set inwhile agricultural run-down is continuing.

Changes in industrial structure imply changes in theoccupational structure of jobs, and these have oc-curred within sectors as well. Table 4 shows rapidgains at the top of the occupational structure withemployment of professional and managerial staffrising on average about 2 per cent faster than totalemployment. Particular countries have experiencedespecially rapid growth of sales workers (USA) orservice workers (Australia), but on average they,along with clerical workers, have maintained arather constant share. Finally the constant absolutenumbers of production workers implies a sharplydeclining share in total employment.

These trends imply an important expansion of op-portunities for those with most educational qualifi-cations and contraction of jobs for those with fewest(Gittleman, 1994). Although the occupational clas-sification is too broad to allow us to draw strongconclusions, these data do not seem to support thenotion that the economy is generating dispropor-tionate numbers of'bad' jobs. Such an impressionmay reflect two other factors. First, more of the badjobs may be visible, behind the hamburger counter


Dow

nloaded from


A. Glyn

Table 3Patterns of Service Employment

Wholesaleand retail trade

Employment shares (%

ECJapanUSAOECD

18.423.522.220.4

Transport andcommunication

. of total), 1992

6.06.05.65.8

Per capita employment growth (average %

ECJapanUSA

0.60.30.9

-0.40.00.2

Finance andbusinessservices

8.510.913.210.4

p.a. 1979-90)

3.32.03.3

Community,social, andpersonal

29.118.531.527.5

(Private only)

2.72.72.2

Generalgovernment

0.6-0.7

0.2

Note: Per capita employment growth is per head of population aged 15-64. Growth rates for community,social, and personal services are for private sector only; remainder are in general government.Source: OECD (19946, Tables 1.1 and 1.3).

Table 4Employment Growth by Occupation, 1980s, Average of Nine Countries

Share (%) early 1990s Average % p.a. changes, 1980s

3.22.91.51.41.5

-1.401.3

Note: Arithmetic average of G7 less Italy plus Australia, Austria, and Belgium.Source: OECD (1994b, Table 1.4).

Professional, technical, etc.Administrative and managerialClerical and relatedSalesService workersAgriculturalTransport, production, and labourersTotal

17.58.7

17.410.210.74.4

30.5100

rather than out of view in sweatshops. Second, thebad jobs have been becoming worse in terms ofconditions of work, levels of pay (relative and evenabsolute), and skill levels. OECD (19946, vol. I, p.163) cites US studies showing that 'new technolo-gies both reduce the skill content and the share oflow-skilljobs while increasing the skill content andthe share of high-skill jobs'.

(iii) Labour-force Growth and Structure

Section II(i) showed that shortage of jobs in theOECD has occurred despite rather steady growth ofemployment. This implies that an accelerated growthof the labour force has been the immediate changebehind the upward trend in joblessness; demand forwork has increased faster rather than there having


Dow

nloaded from



been a slower expansion of available jobs. Thegrowth rate of the labour force in the OECD rosefrom 1.1 per cent peryear during 1960-73 to 1.3 percent per year after 1973; in the EC the accelerationwas more pronounced, from 0.3 per cent per year to0.8 per cent per year.

Population growth does not explain the laboursupply trend; in the USA the growth of populationof working age has slipped and in the EC it has onlybeen fractionally faster than before 1973. The mainfactor has been increased participation in the labourforce. Before 1968, the labour force was growing0.4 per cent per year slower than the population ofworking age in both the OECD as a whole and in theEC; by the 1980s, the labour force was growing 0.3per cent per year faster (0.1 per cent per year in theEC). In the 1960s the average 'participation rate'fell as declining male participation (young menspending longer in education and average retire-ment age declining) outweighed such increases infemale participation as there were.2 But after 1973female participation grew more strongly and since1979 this has offset continued declines in maleparticipation. By 1992 the average participationrate for women was 60 per cent (79 per cent inSweden, very close to the OECD average for men);this represented an increase of one-quarter since1973. As the OECD points out (1994b, Chart 1.9),the rise in women's participation has been closelyrelated across countries to the growth of the serv-ices sector.

Comparing the 1980s with the 1960s, of the 0.6 percent per year acceleration in labour-force growth inthe EC, 0.5 per cent per year is accounted for by thechanging trends in the participation rate. It is thisrise in participation which has meant that even theconsiderable growth of jobs in Europe by historicalstandards has failed since 1973 to match the in-creased numbers of those seeking work. Womenentering the labour force were the most importantsource of demand for additional jobs in services, towhich, of course, they had just as much claim asmen leaving agriculture or industry or young peopleseeking work for the first time. It might be tempting,however, to interpret these trends as implying that

increased labour-force participation was responsi-ble for rising unemployment. But cross-sectionevidence shows the opposite. Although some coun-tries (such as Switzerland, Austria, and Japan) withrelatively good unemployment records had littlechange in labour-force participation, the generalpattern was for countries with the smallest in-creases in unemployment to have substantial risesin participation (Nordic countries and NorthAmerica) while the EC countries with large in-creases in unemployment lagged behind in partici-pation trend (Elmeskov and Pichelmann, 1994,Figure 2).

The average educational level of the OECD poten-tial labour force has been rising steadily; eachdecade's cohort of working age has shown arounda 10 per cent higher share of people with at least an'upper secondary education' than that of the previ-ous decade. The majority of the OECD populationaged between 25 and 64 is now in this position withthe share being more than three-quarters in NorthAmerica, Germany, Norway, and Switzerland(OECD, 19936, Tables C1(A),(B)). Proportions ofthe actual labour force with degrees have also beenrising, frequently by about one-half in the 1980s(OECD, 1994b, Table 1.6). Thus the deteriorationin the relative economic position of the less well-educated, documented in the next section, has oc-curred in the contextof their diminishing numericalimportance in the labour force.

(iv) Conclusion

The jobs deficit in the advanced countries hasreflected a faster growth in the demand for jobssince 1973 rather than slower job creation. Theamount of work provided has, if anything, acceler-ated, but not enough to keep pace with the addi-tional demand for employment arising from in-creased women's participation. Growth rates havedeclined sharply but employment still respondsstrongly to faster output growth. Thus a protractedperiod of faster output growth would lead to asizeable dent in unemployment rates, provided, ofcourse, that the higher level of activity could besustained. The service sector has been the source of

2 In the EC, women's participation was falling until 1968. This reflected the exodus from agriculture where participation ofwomen was sometimes recorded as being very high (in Germany, for example). After 1973 the shift from agriculture slowed (bothrelatively and absolutely) as the agricultural population sank. So the decline in measured female participation rates as a result ofurbanization was less of an influence.


Dow

nloaded from


A. Glyn

additional work and the relatively poor perform-ance of Europe is mainly due to its much greater lossof jobs outside the service sector, and in particularloss of industrial jobs. Relative to the population,however, service employment is not so extensive asin the USA. The occupational structure has shiftedout of manual work, with professional and manage-rial employment growing most rapidly, and therehas been a parallel shift in the educational qualifi-cations of the labour force. This is the backgroundto the employment problems, and in particularincreasing inequality, which have accumulatedwithin the OECD over the past 20 years.

III. LABOUR-MARKET OUTCOMESAND INEQUALITY

A very unequal spread of unemployment is impor-tant both from the point of view of the welfare costsinvolved and because effective policies may needto be focused on the particularly disadvantage*!groups. The main task of this section is to evaluatethe extent to which the unskilled have beenparticularly hard-hit over the past decade and ahalf, a matter of increasing concern in recent years.First, however, some related trends in the distribu-tion of labour-market outcomes will need to beoutlined.

(i) Broad Trends in Labour-market Inequality

Table 5 provides information on various aspects ofrising joblessness in the 1980s. The data are pre-sented for the (broadly) peak-to-peak period 1979-90 in order to focus on the underlying trends (com-prehensive data are not available for the currentrecession, with the effect that the major deteriora-tion in the labour-market situation of Sweden andFinland is left out). Countries are grouped accord-ing to their institutional patterns and responses. Theoverall rise in joblessness is indicated first of all bythe change in overall unemployment (column 1).For the great majority of countries, the recovery inthe second half of the 1980s failed to bring unem-ployment down by as much as it had risen in thepreceding recession. This picture is reinforced by

the data on the change in participation in the labourforce by 'prime age' men.3 The latter measureindicates, for what has been a core section of thelabour force, that the trend in unemployment in the1980s systematically understates the rise in jobless-ness, frequently substantially. In one or two coun-tries (including the UK ) the somewhat reassuringpicture from overall unemployment, that at leastthings did not get worse in the 1980s, cannot besustained once withdrawal from the labour marketis included.

Women and young people (under 25) have beendisproportionately hit by unemployment since 1973.Women's unemployment was typically one and ahalf times the male rate in 1979 (here the UK wasthe leading exception with considerably lower fe-male unemployment). Perhaps not surprisingly inview of the expansion of women's employment,over the 1980s female unemployment typicallyrose less (in absolute terms) than men's, with thespectacular exception of The Netherlands and Spain.Youth unemployment was typically two or threetimes the average in 1979 (Germany being a strik-ing exception). Over the 1980s the pattern is verymixed; the UK and Australia at one end achievedreductions, while a number of EC countries sawyouth unemployment rise much more than the aver-age. In 1979 the share of long-term unemploymentwas less than one-tenth in North America and someScandinavian countries, but 30-40 per cent in mostof the EC. Shares rose substantially in a number ofEuropean countries indicating an increased con-centration of the incidence of unemployment. Fi-nally, earnings inequality, which was higher inNorth America than Europe, and especially North-ern Europe, at the beginning of the 1980s subse-quently rose substantially in a number of countries,most strongly in the UK. It is importantto recognizethat this was not a universal phenomenon; the greatmajority of European countries showed very smallchanges one way or the other and in the case ofGermany, in particular, earnings inequality appearsto have declined.

Country experience in the 1980s falls into roughgroups:

3 Changes in participation rates for women include the strong and relatively autonomous underlying trends discussed above,those of young people reflect growing participation in education, and interpretation of participation by people over 55 is muddiedby the impact of voluntary early retirement These factors are much less important for those in the 25-54 age group.


Dow

nloaded from



Table 5Labour-market Changes, 1979-90 (Changes in Percentage Rates)

JapanCanadaUSAAustraliaUK

Averageunemploymt

00.7

-0.31.20.5

New Zealand 5.5BelgiumDenmarkFranceGermanyIrelandItaly

1.22.33.53.06.93.7

Netherlands 3.4Spain

AustriaFinlandNorwaySweden

7.9

1.2-2.5

3.30.2

LF non-participationmales 25-54

-0.31.81.11.41.9

2.2n.a.n.a.0.93.70.22.4

-0.41.5

0.6n.a.0.5

-0.2

Women'sunemploymt

0.3-0.9-1.4-0.6-0.5

4.50.10.64.03.13.94.26.6

15.5

n.a.-2.7

2.40.1

Youthunemploymt15-24

0.9-0.1-1.6-1.1-2.2

6.2n.a.n.a.5.61.6

10.35.93.0

12.9

n.a.n.a.5.20.1

Share of L-Tunemploymt(%)

50076

n.a.8

-27

1729201225

n.a.n.a.16-2

Wagesinequality(D9/dl)

99

171128

n.a.-2

13

-6n.a.-1-2n.a.

2n.a.-A

1

Notes: Male non-participation in labour force (LF) is % of population aged 25-54.Long-term (L-T) unemployment is in excess of 12 months; changes estimated roughly from graphed datafor 1979.Wages inequality is % change in earnings differential between first and ninth deciles for men except forwage inequality in Denmark and Sweden (men and women).Swedish unemployment and participation rates exclude change in 1986-7 when series break.Date for New Zealand in columns 2 and 4 are for 1986-90, and for Sweden in column 6 are for 1984-90.Sources: OECD Labour Force Statistics, Economic Outlook, Historical Statistics; Freeman and Katz(1994); OECD (1993a, 1994b); Gregg and Machin (1994).

In North America, Australia, and the UK in-creases in unemployment were slight, the rela-tive position of young people and women tendedto improve a little and there was little rise inlong-term unemployment. But there were in-creases in labour-market withdrawal, and earn-ings inequality rose substantially, most notablyin the UK. (Japan's pattern, though from anextremely low starting level of unemployment,is rather similar, though without the labour-market withdrawal.)

In the EC countries (except the UK) unemploymentgrew considerably, along with sharply rising non-participation in Germany and Italy. There weresome spectacular increases in women's unemploy-ment (TheNetherlands and Spain) and youth unem-ployment (Italy, France, Spain, and Ireland) andusually big increases in the share of long-termunemployment. In contrast to the picture of risingjoblessness, with increasingly concentrated inci-dence, earnings inequality changed rather little.(New Zealand resembles the EC countries in terms


Dow

nloaded from


A. Glyn

of the large rise in total and especially youth unem-ployment)

In the Nordic countries, cracks appeared in thegenerally high employment/high wage equalitypattern (Rowthorn, 1992), with rises in unemploy-ment in Norway and in wage inequality in Swedenin the latter part of the 1980s. This was before therecessions in Finland and Sweden destroyed (mostprobably for the medium term, at least) their supe-rior employment record (between the highest pointsfor European unemployment of 1984 and 1994 theunemployment rate in the EC rose by 1 per cent, inSweden by 4.8 per cent, and by 13.1 per cent inFinland).

(ii) The Position of the Unskilled

A number of factors have contributed to increasingattention being devoted to the employment andearnings prospects of the less skilled or qualified.Above all there is awareness of the dangers ofincreasing economic polarization, illustrated by thesituation of the USA and portrayed in RichardFreeman's alarming article later in this issue (seefor example Business Week, 1994, and The Econo-mist, 1994). Second, the thrust of demands for'labour-market flexibility' seems to have shiftedfrom emphasis on the average level of real wages tothe issue of the wage structure and the need forthose at the lower end of the pay distribution to'price themselves into jobs'.

While there are difficulties with any measure of'skill', there is a considerable amount of data avail-able for sections of the population with differentlevels of educational attainment. This corresponds

to part, at least, of economists' notion of humancapital. The OECD recently showed for a numberof countries (19946, Table 7.2) that there is alsoa strong correlation between level of educationand subsequent involvement in training, a find-ing that applies also to the UK (Green, 1994).Recent work for the USA and UK (Juhn et al,1991; Schmitt and Wadsworth, 1994) has docu-mented the extent to which employment variedwith educational attainment, noting in particularthe importance of non-participation in the labourforce. Table 6 provides data for a number ofcountries for the employment rate (ratio of em-ployment to population) of the least educatedgroup (first level of secondary school) relative tothat of the most educated (with university de-grees or equivalent). This, in turn, depends onrelative ratios of employment to the labour force(1 minus the unemployment rate) and relativelabour force participation. Attention is restrictedto men aged 25-54 because, for reasons alreadydiscussed (see footnote 3), the impact ofchanges in the labour-market position of thosewith more and less qualifications shows up moststarkly for this group/

The differences in employment rates are oftenextremely large. With the familiar exceptions ofSweden and Norway, and, less expectedly, Italy,the least educated were 10-15 per cent less likelyto have jobs than the most educated, and in theUSA the differential was 20 per cent.5 Theselevels of disadvantage applied towards the endof the 1980s (1987 or 1988) when labour marketswere not at their most depressed. The importanceof looking at participation as well as recordedunemployment is also underlined;6 inactivity fre-

4 A comprehensive analysis, which is not possible here, would deal in equal detail with women, young people, those over 55,and ethnic minorities. Data for unemployment suggest a broadly similar pattern by educational group for women as for men. InNorth America, the UK, and Australia, but not in Europe, lack of education leads to particularly high unemployment among theyoungest age groups. For the less educated between 55 and 65, participation rates are particularly low (OECD, 1989).

5 Differences between countries have to be interpreted carefully because of the very wide differences in the relative sizes of thetwo groups in different countries. The OECD (1994ft, Table 1.16) made some rough calculations for unemployment rates for thefirst and last quartiles of the labour force ranked by educational qualification for a number of countries. In the case of Germany,the differential between the most- and least-educated categories is considerably more in Table 6 than for the quartiles because theleast-educated group is so small.

' It seems probable that those at the bottom of the educational ladder also constitute a disproportionate number of the'involuntary' part-time workers (put by the OECD at 20-30 per cent of the part-time workers who, as a group, were accountingfor much if not all of increased employment in a number of EC countries) and of the rising numbers of temporary workers in Franceand Spain (OECD, 1994ft).


Dow

nloaded from



Table 6Educational Disparities in Employment 1987, Males Aged 25-55

AustraliaAustriaBelgiumCanadaFinlandGermanyItalyNetherlandsNorwaySwedenUKUSA

Relativeemployment

(2)x(3)

0.880.930.870.830.920.820.940.880.970.980.850.79

Relativejobs/LF

(2)

0.940.960.940.920.940.881.00.950.990.990.880.91

Relativeparticipation

(3)

0.940.960.970.920.980.930.940.920.980.990.960.87

% < fullsecondary

34.723.358.527.071.315.372.344.742.738.644.516.2

% university

13.27.3

18.217.813.912.46.27.2

19.514.019.027.3

Notes: The disparities are between those who have completed less than full secondary education and thosewith higher education qualifications.Relative employment measures the ratio of the employment rates (employment/population) of the least tothe most educated groups.Relative jobs/LF is the ratio of the employment/labour force (100-the unemployment rate) of the least tomost educated groups.Relative participation is the ratio of the participation rates (labour force/population) of the least to mosteducated groups.% < full secondary and % university are the proportions of the population in the age groups with,respectively, less than full secondary education and with university degrees.For Italy the age-group is 25-59.Sources: Calculated from OECD (1989, Tables 2.1, 2A, 2B).

quently contributes as much as unemployment todifferences in the employment rate.7

It may surmised that the position of the least quali-fied deteriorated during the 1980s. Comprehensivedata are sparse but what are available are shown inTable 7. Employment-rate changes are shown to-gether with changes in the relative earnings of thetwo groups. When multiplied together the relativeemployment rates and relative earnings show rela-tive incomes from employment per head of thepopulation group, an index of educational dispari-

ties in income from employment (EDDIE).8 Thetable includes changes in this index.

The declining relative employment position of theless qualified is confirmed (column 1), with sub-stantial worsening in the USA and EC countries.Bigger absolute increases in unemployment aretypical (column 2), and apply to most countries notincluded in the table (for example France, Spain,Canada, and Norway—see Nickell and Bell, thisissue, Table 2). The importance of including par-ticipation changes is underlined. It is striking how

1 Where participation-rate data are not available, the unemployment rates given by Nickell and Bell (this issue) show differencesfor the educational groups in the range 5-9 per cent at the end of the 1980s (France, Spain, New Zealand); only Japan has suchsmall differences as the Nordic countries and Italy in Table 6.

1 It is analogous to Bob Rowthorn 's INDEED index (Rowthom, 1992), combining a country's employment rate with an indexof earnings dispersion across industries.

10


Dow

nloaded from


A. Glyn

Table 7Percentage Changes in Educational Disparities in Income from Employment, Males, 1980s

AustraliaGermanyItalySwedenSwedenBelgiumJapanUKUSA

1982-901978-87

1980-71971-871987-931977-861979-921979-911979-92

Relativeemployment

-3-6-4

1-4-6-2

-12-7

Relativejobs/LF

-1-3-2

0-A-5

0-7-A

Relativeparticipation

-2-3-2

10

-1-2-5-A

Relativeearnings

-33001

n.a.-A

-17-21

EDDIE

-6-3-A

1-3

n.a.-6

-27.-27

Notes: See Table 6. Relative earnings is ratio of earnings of least- to most-educated groups.EDDIE is the product of relative earnings and relative employment rates (see text).Dates refer to the employment data; changes in relative earnings apply to men of all ages and sometimesdiffer slightly in coverage of periods and specification of employment categories from the employmentdata.Age groups for employment are 25-55 for Australia, Belgium, Germany, Sweden, and UK, 25-59 forItaly, and 25-64 for USA and Japan.Sources: Earnings and participation calculated from OECD (1994*, Tables 1.16,7A1, 7B1); unemploy-ment rates and employment rates for USA and UK and Sweden 1987-93 supplied by Brian Bell. Data forBelgium, Italy, and Sweden (1971-87) provided by the OECD.

the change in relative employment rates gives amore pessimistic picture of the position of the leastqualified than does the change in their relativeunemployment rate.9 Some of those dropping out ofthe labour force have moved on to invalidity ben-efit. As BlOndal and Pearson document in theirarticle in this issue, this phenomenon has by nomeans been confined to the UK and in a number ofcountries has been supplemented by quite largenumbers on early retirement schemes. They showthat in some countries the criteria for invaliditybenefit have quite explicitly included the person'schance of finding a job.

When earnings are brought into the picture, thedrastic decline in the position of the least qualifiedin the USA and UK stands out; a fall of one-quarterin their relative income from employment overlittle more than a decade is a huge deterioration. Inthe UK the deterioration in relative employment isalmost as important as relative earnings. This pic-ture of deteriorating job opportunities at the bottomend of the UK jobs market is filled out in graphicdetail in Gregg and Wadsworth's contribution tothis issue. Their evidence suggests a chronic short-age of full-time, reasonably paid, long-term jobs.Unless well-qualified, job-seekers are faced with

' The use of relative unemployment rates, rather than relative employment rates, in most discussions of the problem (OECD,1994M) has tended to suggest relatively little change in the relative position of the least-educated. But it is surely the relativeincidence of the desirable characteristic (i.e. employment) which indicates relative welfare. Consider the relative safety of meansof travel. Suppose the chances of a car accident on a particular journey had gone up from 5 per cent to 50 per cent, whereas thechances of a train accident went up from 1 per cent to 10 per cent The relative chances of having an accident in a car as comparedto a train have remained 5 to 1; but the relative chance of a safe journey by car as compared to train has gone down from 96 percent to 56 per cent The position of car travellers has obviously deteriorated relatively and so does that of the less-qualified if theirrelative chances of a job diminish as compared to the better-qualified, even if the ratio of unemployment rates is unaltered. Thusit is changes in absolute unemployment rate differentials (as used in Table 5 above) or (more comprehensively) in relativeemployment rates (as in Tables 6 and 7) which best show the comparative labour-market position of different groups. This docsnot imply that relative changes in unemployment may not be relevant for wage determination (see Nickell and Bell, this issue).

11


Dow

nloaded from



predominantly part-time, insecure work at pay rateswhich are declining relative to the average and withlittle prospect of this providing a route into a betterjob.

If the calculation for the change in EDDIE in theUSA is taken back to 1975, then the index declinesby a full one-third, almost half of which was due todeclining employment opportunities for the leasteducated (a fall in their employment rate of 15 percent as compared to the best educated). In the USAthe disadvantage was taken mostly in employmentbetween 1975 and 1979 and mostly in earnings after1979. It is very striking that the two countries withthe biggest declines in relative earnings (UK andUSA) also have the biggest declines in relativeemployment, while, conversely, Sweden managedto maintain both, before the rise in unemploymentin the early 1990s which impinged most on the leastqualified. Germany is the one case which combineda severe fall in the relative employment rate of theleast educated with an apparent small improvementin their relative earnings.

(iii) Is There a Jobs/Low Pay Trade-off?

Encouraging 'labour-market flexibility', at the bot-tom end of the pay distribution in particular, hasbecome a major focus of policy advice with weak-ening of minimum wage provisions or changes inthe benefit system being seen as the most fruitfulcandidates for amendment. The evidence presentedin the last section suggests at least that there areroutes other than reducing their relative pay forpreventing a deterioration in employment pros-pects for the less qualified. But the institutions andpolicies which lie behind the success stories are notwell understood and rhay be even harder to repli-cate. So wage flexibility could still seem an attrac-tive option. But does it work?

The OECD presents a chart (1994A, Chart 5.1)which plots the growth of employment over theperiod 1980-9 against the change in wage disper-sion; there would be a strong correlation but for theUK, where the biggest increase in wage dispersionin the OECD was associated with employmentgrowth hardly greater than in other European coun-tries where dispersion hardly rose or even fell. Butworkers in the UK secured much larger than aver-age real wage increases in the 1980s. A better testof the impact of wages at the bottom of the paydistribution may be to look directly at the increasein costs of employing the low-paid. More precisely,the product wage of the low-paid (measuring reallabour cost from the employers' point of view) canbe compared in a rough way with total hours workedin the private sector.10

There is a respectable degree of correlation be-tween the two across OECD countries (Figure2); the regression is significant, with every 1 percent per year slower growth of labour costs forthe low paid being associated with an 0.5 percentper year faster growth of private-sector employ-ment." Inevitably, much of the variability inemployment remains unexplained and must beaccounted for by other factors. Some impact ofreductions in low pay on jobs is consistent withthe evidence of quite high substitutability be-tween unskilled and skilled labour, although asWood (1994) points out, such estimates varyquite widely. Very large impacts seem hard tosquare with evidence from France (Bazen andMartin, 1991), the USA (Card etal., 1993), andthe UK (Gregg et al., 1994) that minimum wageshave small effects (if any) on employment, thoughit may be that monopsonistic structures ratherthan lack of substitution possibilities for un-skilled labour are the explanation. It seems fairto conclude that reducing labour costs at the

10 Real labour cost for the low-paid is estimated from the growth of the average product wage (employee compensation deflatedby the GDP deflator) adjusted for the relative growth of pay in the first as compared to the fifth decile, using the figure for menand women combined where possible (OECD, 1993a, Table 5.2). It is clearer to compare hourly pay with totaJ hours worked, forthe employment figures are affected particularly by the growth of part-time work which is a distinct issue. Focus is on the bottomof the pay distribution since few have argued that increased dispersion at the top of the earnings distribution creates more jobs!In both the US and UK more of the increased wage dispersion has resulted from rising differentials at the top than at the bottomof the distribution.

" If the growth of average product wages is substituted for low-paid product wages the regression with employment growthis wholly insignificant. There has been a very genera) increase in profitability in the 1980s (product wages growing slower thanproductivity), but whatever positive effect that might have on employment has been dwarfed by causation running in the reversedirection—the more unemployment rose, the slower product wages grew relative to productivity (Glyn, 1994).

12


Dow

nloaded from


A. Glyn

Figure 2Employment and Low Pay, 1979-90

cdO.

o

cd

uo

3O

JZ

o

2 -

1 -

-1 -

-2

•AUSL

4

U!

-

sv

-

IA

\/ED

CAN

•NETHS

i

UK

•DEN

•AUST

JAP

GER•

NOR

BEL

FRA

i

4IT

i

•FIN

-1 1 2Real low pay labour costs (% p.a.)

bottom end will have some effect on employ-ment, but just how substantial is very uncertain.

(iv) Conclusion

During the 1980s most countries displayed eitherincreased joblessness, with more long-term unem-ployment and more concentration on women oryouths (many EC countries), or greater earningsinequality (the Anglo-Saxon economies, Japan).Even the high employment/egalitarian earnings dis-tribution typical of the Nordic countries was fray-ing. The disadvantaged position of the least-edu-cated was not confined to pay; in many countries thechances of their having work were much smallerthan for the best-educated, especially when fallingparticipation is included. In the UK and USA therewas a dramatic worsening of their position in the1980s, and some deterioration was apparent in mostcountries. OECD experience in the 1980s suggeststhat falling relative wages for the least-qualifiedwas certainly not a necessary condition for theiremployment rates to be maintained; there was sometendency, however, for employment to rise fastest

where labour costs at the bottom of the pay distribu-tion grew most slowly.

IV. EXPLANATIONS FOR GROWINGLABOUR-MARKET INEQUALITY

It is rather easy to rule out the most obvious supply-side explanation for the declining job and earningprospects of the least-qualified. Since their share inthe labour force has been declining (as shownabove) it cannot be put down to their increasingnumbers. However, even if declining, the numbersof the least-qualified may not be falling as fast aspreviously in the face of long-term trends in thepattern of demand. Such an explanation (in reverse)has been suggested for rising earnings differentialsfor college graduates in the USA and the slow-down in the rising share of university graduates isa rather widespread phenomenon according to theOECD. A glance at the pattern of educationalattainment over time suggests that it m ight be worthconducting a systematic analysis of the numbers ofless-educated as well. In a number of countries

13


Dow

nloaded from


OXFORD REVIEW OF ECONOMIC POLICY, VOL. 11. NO. 1

where upper secondary-school education had ex-tended to very large percentages of the populationsome decades ago (North America, Germany, Nor-way, Switzerland) the cohort aged 25—34 shows nofurther increases in educational attainment (and, inthe case of the USA, apparently a slight decline(OECD, 19936, Table Cl)). Even if these trendsprove to be important, the weakening in demand forless-qualified labour remains the central underly-ing factor. This will be discussed next.

It is helpful here to distinguish between the effectsof sharp reductions in demand and longer-termchanges. The former should not be neglected. Therehave been three big upward lurches in unemploy-ment (Figure 1) which have not been reversed. Thelabour-market position of the least-educated dete-riorated in absolute terms when total unemploy-ment rose (Nickell and Bell in this issue, Table 2).The recent rise in unemployment in Sweden is agood example. The difference in the unemploy-ment rates of the least- and most-educated rosefrom 1.5 percentage points at the end of the 1980sto 6 percentage points in 1993 as unemployment onaverage rose by 7 percentage points. Either demandreductions were concentrated on industries wherethe least-qualified are employed, and/or firms werehoarding their more qualified labour. But sucheffects would presumably be temporary, as labourhoarding was unwound and demand patterns re-turned to normal, if there were no longer-termchanges in the structure of demand for labour.12

(i) Trade and Technology

The most debated issue is the role of trade with theSouth in hastening the replacement of low-skilledjobs in the North by those demanding higher educa-tional requirements. At first sight it seems mostimplausible that imports of manufactures from low-wage countries could have played a very importantrole. After all, they were only a tiny fraction of themarket for manufactures in the North (4.3 per centof the domestic market in the USA in the secondhalf of the 1980s, around 3.1 per cent in the big ECcountries, and 2.6 per cent in Japan (Martins, 1993));

they were also, by and large, balanced by exportsfrom the North to the South. In his book AdrianWood (1994) argues that these imports would re-quire very much more labour to produce in theNorth than their low value would suggest. Hisestimates suggest that manufactured imports fromthe South would have been nearly three times asexpensive if produced in the North (Wood, 1994,Table 4.5). Correspondingly, very many more jobswere lost in the North than were gained in the moreskill-intensive sectors which produced the goods(machinery, etc.) exported to the South to pay forthe labour-intensive imports. Wood goes on toclaim that this direct effect has been compoundedby the pressure which the imports have brought tobear in forcing the rationalization (and thus job-shedding) in Northern industry.

Wood's fundamental point concerns the relativedemand for skilled versus unskilled (or more- andless-educated) labour. Both falling relative pricesof imports from the South, and the defensive pro-ductivity growth they may engender, increase realincomes in the North. Thus, assuming permissivemacroeconomic conditions, there will be large off-setting demand and employment increases in othersectors. Wood's case depends on the employmentthereby generated being, on average, of higher skillintensity than the jobs lost in labour-intensive manu-facturing. Further, for there then to be an importanteffect on total unemployment, it must be the casethat the twisting of demand away from unskilledlabour reduces the sustainable level of employmentoverall. For example, this could be because theeconomy runs into skill bottlenecks, and thus risinginflation, at a higher overall level of joblessness.Southern trade would then be the generator of anincreased degree of labour-market mismatch, whichwould in turn push up the NAIRU, or prevent itfalling to the extent indicated by other develop-ments in the 1980s.

This argument has provoked a heated controversy,particularly in the USA, where imports from theSouth have been largest, where the position of theleast-educated has deteriorated very drastically as

12 OECD (19946, p. 13) reports US studies which point to a proportion of workers being employed in jobs for which they are'over-qualified', but not to an increase in this proportion which might explain the deteriorating position of the least qualified bytheir being 'bumped down' off the end of the jobs ladder. It is important to discover whether this is also true of European countrieswhere the trend in unemployment has been sharply upward.

14


Dow

nloaded from


A. Glyn

described earlier, and where the debate about theNorth Atlantic Free Trade Area (NAFTA) hasgiven the whole issue an unusual degree of policyrelevance. A recent, very thorough analysis pro-vides at least partial support for Wood, concludingfor the US that,

As a result of increased internationalization, employ-ment has declined sharply in low-skill sectors and hasincreased in high-skill sectors. In addition, the increasedtrade has contributed to falling relative prices of lessskill-intensive goods and to the growing inequality ofearnings between low-skilled and high-skilled workers,although the weight of the trade effect is uncertain.(Sachs and Shatz, 1994, p. 57)

The OECD was much more doubtful. Significant,though quantitatively small, effects ofNorth-Southtrade in reducingNorthern employment were foundin a 'small number of specific industries (textiles,clothing, footwear, computers, radio-TV-commu-nication equipment)'; however, it concluded, sur-prisingly, that this effect was 'most noticeable inindustries employing a high proportion of skilledworkers'(19946, p. 104).

This debate will clearly stretch to many more arti-cles and it is too early to predict where the consen-sus will settle. But trade explanations will, at thevery least, have to be complemented by acceptanceof an important role for technical progress. In thevery simplest version of the trade story, competi-tion from the South wipes out Northern sectorswhich use a great deal of less-qualified labour,leading to some combination of falling relativewages and rising relative unemployment for theunskilled. It is an example of the pattern of produc-tion moving against the unskilled. But the pricesystem should generate some offsetting effects. Tothe extent that the relative wages of the unskilleddecline then unskilled labour should be replacingskilled labour within industries. However Sachsand Shatz's data (1994, Table 11) show that therewas a major reduction in the share of productionworkers (and, they conclude, a corresponding risein the skill intensity of production) within the wholerange of manufacturing industries. This evidence isthe more impressive because it refers to the USA,where the relative wages of the unskilled fell somarkedly. A direct link between declining propor-tions of unskilled workers within industries and

technology has been made by studies showing thatthe sh ift towards non-manual employment occurredmore rapidly in those manufacturing industriesidentified as introducing new technologies (Bermanetal., 1993,fortheUS;Machin, 1994, for the UK).

(ii) What about the NAIRU?

A decade ago, when unemployment was last ana-lysed in this Review, the NAIRU model was takingover as the dominant paradigm for understandingEuropean unemployment (see Rowthorn, 1977;Carlin and Soskice, 1990; Layard et al., 1991).Particular stress was laid on increased trade-unionpower as a result of the extended period of highemployment, more powerful organization, and thereduced fear of job loss implied by higher unem-ployment benefits. The resulting attempts to gainimprovements in living standards and extensions tothe welfare state collided, after 1973, with deterior-ations in the terms of trade as a result of the OPECprice increases .

Numerous estimates have been made of the NAIRUfor individual countries, that is of the 'equilibrium'rate of unemployment at which the inflation rate hasno tendency to rise or fall. As elaborated and testedin Layard et al. (1991), the pattern of unemploy-ment is related to variations in the pressures im-posed on the bargaining system (for example, realincome cuts as a result of terms-of-trade deterior-ations) and to differences in the bargaining struc-ture which had to cope with these pressures.

Bargaining systems are differentiated according towhether wage determ ination operates at the level ofindividual firms or plants at one extreme or acrossmore or less the whole economy at the other, withindustry-level bargaining occupying the middle ofthe spectrum. The location of bargaining affects themarket power of the participants and the extent towhich they themselves bear all the consequences oftheir actions. A whole industry, if it is protectedfrom international competition, has more latitudefor raising prices than an individual firm. Thus partof a wage increase will be paid for by higher prices,and thus lower real incomes, in the rest of theeconomy. Atthefirm level, by contrast, moreof anymoney-wage increase will represent higher realcost of employing labour for the firm, with conse-

15


Dow

nloaded from



quent effects on employment. If bargaining appliesto the whole economy, then money-wage increaseswill lead to proportional price increases with nobenefit to the workers (or if profits were squeezedjobs would be lost). Bargaining at the national levelmeans that the externalities of the bargain areinternalized. Calmfors' extremely clear and com-prehensive review of the whole issue (1994) dis-cusses the full range of externalities, of which priceincreases is merely the most obvious. Based on thiskind of reasoning, Calmfors and Driffill (1988)derived their famous hump-shaped curve to suggestthat bargaining would be most aggressive, and thusreal wages and unemployment highest, in econo-mies with intermediate levels of bargaining.

One important issue, emphasized in Soskice (1990),was that it was the actual level at which coordina-tion of wage-setting was carried out which wasdecisive, not the formal location of wage bargain-ing. Thus the Japanese system of firm-level bar-gaining could still be highly coordinated if firms, orunions or both closely coordinated their bargainingstrategies, taking into account the impact on theeconomy as a whole; alternatively, a governmentincomes policy could lay down general norms withinwhich bargaining would take place at a much morelocal level. It is difficult enough to define in aprecise way even the formal level of centralizationof bargaining. When it is replaced by the degree ofcoordination, disputes are multiplied about where aparticular country's system sits on the scale and (aneglected issue) how this changes over time.

The argument has been extended in a number ofimportant ways. For example, industries subject tointernational competition are thereby relegated, interms of market power, to something closer to theposition of individual firms, flattening the 'hump'.Calmfors (1994) conjectures that the net result of anumber of such considerations is that the humpsurvives but that extreme centralization could beexpected to produce the best outcomes. But eventhis tentative result is subject to a further complica-tion when national bargaining is supplemented bybargaining at an industry and/or firm level (as hasbeen argued has always applied in the Nordiccountries). In such circumstances, the need forbargainers at all levels to demonstrate their capacity

to achieve money-wage increases may mean thatrestraint of average real wages, together with nec-essary changes in relative wages, can only be ac-commodated within a relatively high inflation rate.Such a very Keynesian role for changes in the pricelevel could mean that, at low inflation, national (inreality multi-level) bargaining may generate worseemployment outcomes than decentralized or evenindustry bargaining.

It is perhaps not surprising that the empirical workhas yielded ambiguous results. Although nationalbargaining seems to score fairly consistently overindustry bargaining, there is no consensus on theexistence of a hump. Not only are the explanatoryvariables difficult to measure, but the number ofcountries is small, so that rearrangements of a fewrankings can have a big effect. Since, by and large,they all had 'full employment' in the 1960s and1970s, what is being examined is the ability ofdifferent bargaining systems to withstand the'shocks' of the early 1970s and 1980s. Calmforsconcludes that 'in the end we do not have muchmore than individual country examples that may beopen to many interpretations' (1994, p. 182). Onefairly robust result (Layard et al., 1991; Bean,1994) seems to be that employer coordination hasbeen more important in securing wage moderationthan union coordination (while the theory concen-trated on unions).

Most of the empirical work stops before the latestrecession. The spectacular rise in Finnish unem-ployment in the early 1990s, from one of the lowestto one of the highest in the OECD (together with thesmaller rise in Sweden), has eliminated the veryfavourable employment record of the highly cen-tralized group. But the Finnish economy, in particu-lar, was hugely and uniquely affected by the col-lapse of trade with the former USSR (and Swedensuffered particularly severe consequences fromheadlong financial liberalization). No degree ofbargaining flexibility could have prevented fiercerecessions. It is notable, in addition, that in bothcountries moves to decentralize bargaining werereversed to deal with the crisis (Vartiainen, 1994).It would seem unwise to draw strong conclusionsabout bargaining structures from the high unem-ployment in these countries.

16


Dow

nloaded from


Unemployment is not the only aspect of labour-market performance which may be affected by thebargaining system. Wage dispersion seems defi-nitely to have been lowest in the countries withcentralized bargaining systems, such as Swedenand Norway, with the USA having very high differ-entials (Rowthorn, 1992). The view that relativewage flexibility (to generate jobs for the least-qualified) is assuming equal importance with ag-gregate wage flexibility (to generate jobs in gen-eral) is strengthening the hand of those arguing fordecentralization and deregulation of bargaining.But the case of Austria (very centralized bargainingbut with high wage dispersion) shows how difficultit is to draw simple conclusions. Calmfors' viewthat 'any proposals for change must build on exist-ing institutions and traditions' (1993, p. 182) ishighly pertinent to the drawing of policy conclu-sions.

If the simple NAIRU approach sketched out aboveis adhered to literally, then the structural trendsoutlined in sections II and in of this overview haveno bearing on unemployment outcomes in total,only on which groups of workers are most affected.The fact that male industrial workers were losingtheir jobs while women previously out of the labourforce were starting work in the services, would leadto the pool of unemployed consisting dispropor-tionately of male, manual, ex-industrial workers.But the size of the pool would be determined onlyby the extent to which total unemployment had torise to keep inflation from increasing. This woulddepend on terms-of-trade deteriorations, bargain-ing structures, and so forth. But for structural changeto be irrelevant, wage pressure must depend only onthe total level of unemployment, without regard towhat type of workers were unemployed, from whichindustries, in what part of the country, and for howlong.

Unease with such a conclusion grew with the expe-rience of the 1980s. Trade unions suffered majorset-backs in Europe and the USA (the Fiat dispute

A.Gtyn

in 1979 in Italy, the UK miners' strike of 1984/5,and the air traffic controllers' strike in the USAbeing important examples). Union membership hasdeclined in a number of countries, as has the cover-age of collective bargaining (OECD, \994d). Anumber of countries including the UK toughenedthe unemployed benefit system in the 1980s, re-versing the trend of the previous two decades (seeBldndal and Pearson, this issue) and minimumwages were frequently reduced or even abolished,which should have further weakened the bargain-ing stance. Moreover, the deterioration in the termsof trade of the early 1980s had been reversed by1990 (easing the pressure on real wages) and in-creases in taxation were more moderate as welfare-state spending was held back. Thus the overallthrust of developments on the supply side shouldhave generated some reversal in the upward trend ofthe NAIRU.

Yet Barrell et al. (1994) found that, of the big fourEuropean countries, the only shift down in theNATRU occurred in Italy (reflecting in part thedismantling of wage indexation). In all these coun-tries NAIRU estimates were still in the range 8-10per cent by the end of the decade. The OECD(1994fc, Chart 2.4) found only a slight, and suspi-ciously temporary, decline in the EC NAIRU at theend of the 1980s. A very detailed analysis of the UKlabour market, which above all others should haveregistered the impact of weaker trade unions andlabour-market deregulation, strongly confirmed thebroad findingsoflittlechangeintheNAIRU: 'Mostauthors agree that the Thatcher reforms have hadvery little effect on unemployment or wage forma-tion' (Barrell, 1994, p. 13).13 The OECD's latestNAIRU estimates are shown in Table 8. The onlybig shift is the non-EC European NAIRU shootingup in the early 1990s, another example of theNAIRU trailing along behind the trend in actualunemployment1*

In this connection BlOndal and Pearson's detailedexamination of the impact of unemployment and

13 Bentolila and Dolado (1994) argue from the Spanish experience that labour-market deregulation may paradoxicallystrengthen the position of core workers by creating a protective layer of temporary workers.

M Actual unemployment rose in Sweden because the gyrations of financial and macroeconomic policy engendered a 15percentage point rise in the savings ratio between 1989 and 1993 and a 40 per cent fall in business investment Whatever the longer-term problems with the Swedish model (see Calmfors, 1993; Meidner, 1993) they could certainly not explain a sudden leap inthe NAIRU. The rise in the estimated NAIRU just expresses the smaller effect on inflationary pressure of the actual increase inunemployment than previous experience suggested.

17


Dow

nloaded from



Table 8OECD Unemployment and Estimated NAIRUs, 1973-94, Average Percentage Rates

Non-Europe—unemploymentNAIRU

Big 4 Europe—unemploymentNAIRU

Smaller EC—unemploymentNAIRU

Other Europe—unemploymentNAIRU

1973-9

4.64.84.34.34.74.61.31.3

1980-9

5.95.88.78.4

13.412.52.12.1

1990-3

5.65.48.89.0

12.914.13.83.5

1994

5.75.5

10.29.4

16.715.05.85.1

Notes: For Non-Europe, the first column refers to 1970-9. The OECD describes the NAIRU as NAWRU(non-accelerating wage inflation rate of unemployment) which is 'the unemployment rate below whichwage pressures start to emerge'.Source: OECD (1994c, Table 12).

other forms of state benefit is very relevant. Incommon with some earlier studies, they find thatthe generosity of unemployment benefits has someimpact on recorded unemployment. But they alsofind just as big an effect on labour-force participa-tion, so that higher benefits are not associated withlower employmentrates. Conversely, high sicknessand invalidity benefits reduce labour-force partici-pation. On this evidence the structure of state ben-efits appears to affect mainly the relative size of thevarious categories of the non-employed rather thanthe total number in work.

The NAIRU approach has been developed by tak-ing into account differences in impact on wagebargaining of different sections of the unemployed.The long-term unemployed, it is suggested, exertless of a downward pressure on wage bargainingthan those who have recently lost their jobs. This isbecause their capacity to work may be diminishedby erosion of skills or may be thought to be low byemployers who take long-term unemployment as asignal of low ability. One way or another, theirability to compete in the labour market is reduced(Layard et al., 1991). The implication is that theeffect of increased unemployment in holding downwages erodes as the proportion of long-term unem-ployed grows.

In the extreme case of all the additional unem-ployed becoming long-term and all of them ceasingto function as part of the reserve army of labour,then a higher level of unemployment would ceaseafter a time to have any impact on wage bargaining(as in the 'insider-outsider' models). Such purehysteresis is an extreme position and would seemhard to reconcile with the microeconometric evi-dence that relative wages within an economy areaffected by local unemployment rates according toa 'wage curve' (Blanchflower and Oswald, 1994).More plausibly, some of a rise in unemploymentwill take the form of long-term unemployment sothat its impact on wage bargaining diminishes overtime (but does not disappear).15 A recent OECDstudy confirmed that there was generally a strongereffect of short-term unemployment on wages thanlong-term, but that long-term unemployment didhave some impact (typically one-third of the effectof short-term unemployment (Elmeskov andMacFarlan, 1994)). The UK was one of the coun-tries where the long-term unemployed apparentlyexerted no depressing effect on wages at all, whichmight suggest particularly severe detachment fromthe labour market. It is noteworthy, however, that inthe unusually sharp recovery of employment in theUK in the later 1980s, long-term unemploymentdeclined as fast as short-term unemployment, even

15 In this a case a shock leads to a larger increase in the NAIRU in the long-term (when long-term unemployment has risen) thanin the short-term (when all additional unemployment is of people who have recently lost their jobs)—see Carlin and Soskice(1990). Any structural change which increased the share of long-term unemployed for given total unemployment, as might resultfrom job loss becoming more concentrated on the least qualified, would increase the NAIRU.

18


Dow

nloaded from


A. Gtyn

adjusting for the impact of the government's Re-start programme (Robinson, 1994).

Such an elaboration of the NAIRU approach isquite consistent with emphasis on the pattern ofdemand for unqualified labour or any other factorwhich causes unemployment to be concentrated onparticular groups, such as young people or women.For the particular form that a rise in unemploymenttakes may well affect the extent to which long-termunemployment rises and thus the eventual level of'equilibrium' unemployment. For example, sup-pose large numbersof comparatively well-paid, butnot highly qualified, male industrial workers arelosing their jobs in geographically concentratedwaves of plant closures, and that many new jobs arein service sectors which are accustomed to payinglower wages to mainly women, often part-time,workers. This would set the scene for a large in-crease in long-term unemployment as those maderedundant failed to find what they regard as 'de-cent' or acceptable jobs; many may end up 'with-drawing' from the labour market entirely (thoughperhaps only temporarily).

Indices of 'mismatch', based on the variance ofunemployment rates across skill categories or re-gions, do not show much increase in the 1980s.There was initially some tendency to interpret thatas showing that mismatch was not important, inapparent conflict with the cross-section finding forUS states (Summers, 1986) and for OECD coun-tries (Glyn and Rowthorn, 1988) that the degree towhich unemployment had risen was closely corre-lated with the degree to which industrial jobs hadbeen lost It has been pointed out by Layard et al.(1991, ch. 6) that even constant shares of mismatchin total unemployment would leave mismatch ex-plaining perhaps one-third of the overall increasein unemployment on their estimates. Inclusion ofinactivity, which as discussed earlier is heavilyconcentrated on the less qualified, could also in-crease the weight of mismatch. As explained inLayard et al. (1991) the appropriate indicator ofmismatch depends on exactly how inequality ofunemployment rates affects wage bargaining andthus the NAIRU. Wood's (1994) approach, whichemphasizes the key role of skilled labour in wagebargaining, suggests that all the excess unemploy-ment of the less-skilled as compared to the skilled

represents mismatch. Thus most of the increase inthe average unemployment rate, that is all theexcess rise compared to the skilled rate, would bethe result of the relative decline in demand forunskilled labour. This debate has yet to be resolved(see Abraham, 1991), but it may very well be that ashift in the pattern of demand towards more quali-fied workers has been an important factor prevent-ing supply-side changes from pushing the NAIRUdown.

The 'industrial reserve army' is obviously an ex-tremely complicated phenomenon, as Marx pointedout, with varying degrees of displacement from thecore of stable employment—temporary or casualemployment, short-term unemployment, long-termunemployment, economic 'inactivity', and so forth.It is hard to imagine that the size of the variouspools, and movements of individuals between them,is independent of structural change within theeconomy and that all aspects relevant to wagebargaining can be captured by one number, be ittotal unemployment, or short-term unemployment.From this perspective, a stable relationship be-tween one indicator of the labour-market condi-tions and wage increases would seem most surpris-ing. This does not imply that labour-market condi-tions do not matter and that a rise in overall unem-ployment will have no effect on inflation, only thatits impact at any particular time is very hard topredict.

(iii) Inequality as an Equilibrating Mechanism

The assumption thus far has been that the deterio-rating labour-market position of the least-qualifiedreflects some twist in the balance of labour demandaway from them at a rate which has exceeded thefall in their share of the potential labour force.Nickell and Bell in this issue conclude that this is farfrom the whole story and that much of the greaterabsolute rise in their unemployment rate is theresult of a complex chain of labour-market adjust-ments to a general fall in demand which is not, in thefirst instance, biased against them. Their argument,in brief, is that a general fall in demand, leading toequal rises in the percentage unemployment rate forthe two groups, will have a smaller impact on thewages of the unskilled, who will, therefore, sufferfurther job loss as skilled workers become rela-

19


Dow

nloaded from



tively cheaper to employ. The reason that the wagesof the unskilled would not decline as much is that anequal percentage point rise in unemployment repre-sents a proportionately smaller rise in unemploy-ment for the unskilled. This is because they have,for a variety of reasons, a lower initial level ofunemployment. If it is the proportionate change inunemployment which affects wages, and the evi-dence for this is discussed in the paper, then thewages of the unskilled will be less affected. De-pending on the degree of substitutability betweenunskilled labour and skilled labour, the rise inrelative wages of the unskilled will lead to a secondround of job losses for them, while the loss of jobsfor the skilled is cushioned to the extent thatthey arenow substituted for the unskilled. If the degree ofsubstitutability between the two types of labour ishigh, then the final outcome could be nearly equalproportionate increases in unemployment for thetwo groups, far removed from the equal absoluteincreases which was the starting point.

Nickell and Bell accept that this cumulative disad-vantage for the less qualified—their high initialunemployment rate dooms them to bigger in-creases—has been exacerbated by a non-neutralshift in labour demand way from the unskilled, beit from trade or technology. The very fact that therelative wages of the unskilled have fallen sharplyin a number of countries shows that changes in thepattern of demand, and not just its overall level,must have been very important (since general de-creases in demand are predicted by them to raise thewages of the unskilled). But, conversely, Nickelland Bell argue that a declining overall level ofdemand for labour has to be accorded a central role.Leaving aside their detailed results, which obvi-ously depend on estimates of a number of param-eters, the general point is that understanding thecauses of rising labour-market inequality, and thusdevising remedies, must be based on a full analysisof labour-market responses and not just on impacteffects.

(iv) Capital Stock Growth

Another feature of the 1980s has been the relativelyslow rates of capital accumulation (3 per cent peryear capital stock growth in Europe since 1973) Onsome accounts this has no implication for the

NAIRU, but Rowthorn's article in this issue sug-gests three routes through which low investmentmay have exacerbated unemployment. A higherlevel of capacity utilization may reduce real wagesthrough relaxing the competitive pressure on firms,thus allowing them to secure higher profit margins.It may also reduce real wages because lack ofcapacity worsens the overseas payments positionand thus requires a lower real exchange rate toachieve balance. In order to prevent wage pressure,unemployment will have to be higher. In addition,slower growth of the capital stock will reduceproductivity growth and higher unemployment maythen be required to bring the trend of wages intoline. These effects of weaker investment may haveoffset other labour-market trends that would havebeen expected to reduce the NAIRU (weaker un-ions, for example). The fact that fewer 'good jobs'will be generated in manufacturing and businessservices when manufacturing investment is lowwill then impinge on the less-qualified, who facemore competition for less well paid jobs, in con-sumer services, for example. Rowthorn's generalargument is that both overall unemployment, andits differential incidence, may reflect broader as-pects of macroeconomic development rather thanjust labour-market characteristics. The implicationis that higher investment can play an important rolein employment creation.

(v) Conclusion

The connection between the deteriorating relativeposition of the less-qualified and overall jobless-ness is complicated. In the argument ofNickell andBell, much larger falls in unemployment for theunskilled could be generated by factors which ini-tially increase unemployment across the board.Nothing extraneous by way of structural shiftstowards skill-intensive industries or technologies isrequired. The apparent upward drift in the NAIRUin the 1980s, despite changes in the position of tradeunions and benefit systems which would tend tohave reduced it, still needs explaining. But anyfactor which raised the NAIRU would tend to havea larger absolute effect on the unemployment rate ofthe unskilled. The alternative approach sees changesin the structure of trade or in technologies as havingspeeded up the switch away from unqualified la-bour and the resulting concentration of joblessness

20


Dow

nloaded from


A. Glyn

as having increased the NAIRU or prevented itfalling as skill shortages appear at lower levels ofunemployment. The two approaches are not incon-sistent, but their relative importance is contentiousand debate on the matter is in full swing. Considera-tion of appropriate policies need not await its reso-lution.

V. POLICIES TO REDUCEUNEMPLOYMENT

Snower's article in this issue presents a systematicoverview of unemployment policies. He argues thatthey should be evaluated in relation to the empiricalrelevance of the theoretical ideas which underpinthem and places particular emphasis on policieswhich reduce the costs to firms of labour turnover.In line with the theme of this overview, the focushere will be on policies aimed at the deterioratingposition of the less-qualified and on how differentpolicy approaches imply a particular distribution ofthe costs of reducing unemployment.

The issue of inflation cannot be ducked just becausethere is a structural element (more- versus less-qualified workers) involved. A useful bench-markis a generalized demand expansion of the tradi-tional Keynesian type. It was pointed out above thatthe relative employment position of the least-quali-fied tends to deteriorate in recessions; conversely,upswings in demand improve their position. Lackof demand has certainly exacerbated the impact ofthe more fundamental trends which have been work-ing against them. A strong upswing would obvi-ously help. The basic problem is the NAIRU.

According to current OECD estimates (reproducedin Table 8), actual unemployment was close to theNAIRU in North America, nearly 1 per cent abovethe NAIRU in the four major EC countries and non-EC Europe, and nearly 2 per cent higher in thesmaller EC countries. On this basis, very little of thecurrent level of unemployment is attributed to gov-ernment attempts to reduce inflation by pushingunemployment above the NAIRU. Conversely, ageneral demand expansion would soon generateinflationary pressures and only a small reduction inunemployment could be sustained for more than afew years.

As pointed out earlier, on some versions of theNAIRU model, expansions would have the sameimpact on inflation regardless of whether they wereconsumer- or investment-led. If investment is ac-corded a moderating effect on inflationary pres-sure, as Rowthorn suggests, then attempting toensure that a recovery is investment led is ex-tremely important (a position strongly endorsed bythe Commission of the EC (1993) which put in-creasing the share of investment at the heart of itsrecovery programme). Even over and above thedifficulties of engineering an investment expan-sion, a further problem derives from the financialmarkets. Fearing the worst, they can undermineexpansions as the currency depreciates and/or in-terest rates rise. Even coordinated expansion at theEuropean level, often advocated as a way to circum-vent the balance-of-payments and exchange-rateeffects of unilateral reflation, only avoids one cor-ner of the inflationary problem and leaves intact thedomestically generated elements.

The position of the least-qualified may be tackledmore directly, either on the supply or demand side(for an extremely helpful review see Wood, 1994,ch. 10 ). If the NAIRU is, in part, a function of skillshortages, then policies to enhance the skills of theunemployed offer a way of reducing it. To take thecrudest example, if wage pressure depended onlyon the unemployment rate of the skilled, then train-ing unemployed workers so that they became skilledwould reduce the NAIRU. Provided the govern-ment ensured that demand expanded appropriately,then the newly skilled workers could find jobs withno impact on wage pressure. The effect would beeven greater as the expanding employment of theskilled would, given permissive demand condi-tions, also generate complementary unskilled jobs,again without inflationary pressure. Such an ap-proach is egalitarian in terms of its effects on bothemployment and the distribution of earnings; fur-ther, it will tend to enhance long-run productivecapacity via its effects on the stock of humancapital. Assuming the budget was already appropri-ately balanced, the costs of training (includingemployment of trainers and extra allowances forthe unemployed) would have to be met out ofgeneral taxation and the cost per additional long-term job created would depend on the effectivenessof the programme in enhancing skills and the im-

21


Dow

nloaded from



pact of additional supply of skills on the NAIRU.The design of such programmes raises a host ofissues (see Oxford Economic Papers, 1994, forexample). The results of the detailed study byCalmfors and Skedinger of Swedish Active LabourMarket Policies later in this issue are rather encour-aging as to the impact of the training programmesimplemented there in reducing unemployment.

Given limited possibilities for the unskilled andjobless to be trained into being skilled and em-ployed, attention has also been devoted to boostingthe demand for unskilled labour. The market ap-proach to encouraging such demand is to ensurethat the cost of unskilled labour falls by removingimpediments to such flexibility in the form ofunemployment benefits or minimum wages. Someevidence for supposing that the behaviour of pay atthe bottom of the scale has affected the relativespeed of job creation across countries was givenabove. The relative underdevelopment of Europeanservices as compared to the USA noted earliersuggests one route through which pay flexibility atthe bottom end might promote employment crea-tion. But available evidence suggests that the priceelasticity of demand for services, while consider-able for individual services, is very small for serv-ices as a whole (Summers, 1985). Thus reducingtheir price, via cuts in low pay, does not appear tobe a very promising route to higher employment. Ifskilled and unskilled labour can be very readilysubstituted, then a relatively small fall in the rela-tive wages of the unskilled would provide a way ofmaintaining demand for the less-qualified in theface of underlying trends in the opposite direction.But there are widely differing interpretations of theavailable evidence on this (see Nickell and Bell,below, and Wood, 1994) and it seems hard toreconcile the very large declines in both the relativedemand for less-qualified labour and in their rela-tive wages, at least in the UK and USA, with a veryhigh degree of substitutability. If substitutability isnot high then it is clear that the distributionalimplications of relying on wage flexibility are highlyinegal itarian—the worse-paid sections of the popu-lation have to bear the cost of reducing unemploy-ment via substantial cuts in their wages while the

better-off sections of society benefit from the cheaperservices.16

Concern at the distributional implications of suchpolicies has led to suggestions (Dreze andMalinvaud, 1994) that instead of encouraging de-clines in relative wages, the employment of the less-qualified should be subsidized via their employerspaying lower social security contributions (as isalready the case to some degree in the UK) or moresubstantially via direct employment subsidies ofone sort or another. The Commission of the EC(1993, p. 142) reports simulations that reductions insocial-security contributions at the bottom end ofthe pay scale, financed by a carbon tax, could havequite substantial effects on employment. An alter-native tack accepts that market pressures have toforce down relative wages (and thus labour costs toemployers) but aims to support the incomes of theworkers concerned via some form of negative in-come tax for the low paid, or most radically by abasic income for everybody. The starting point forsuch proposals is the belief that the problem ofunemployment should not be solved at the expenseof the living standards of the least well off, butrather that such costs should be broadly spread. Theproblem, of course, is that there has to be socialacceptance that such costs should be met; otherwisethe NAIRU will be pushed up as wage bargainingattempts to compensate for the extra taxation re-quired to fund the employment policies.

The final approach is the direct public-sector crea-tion of jobs, which can be concentrated on sectorswhere less-qualified workers predominate (con-struction or some basic public services would beexamples). It is strikingthatthe Scandinavian coun-tries, where public-sector job creation played suchan important role, managed best to preserve theemployment and earnings position of the least-qualified. The impact on employment of such gov-ernment spending is much more predictable thanthe impact of wage-cutting or subsidization of em-ployers or workers in private services. No cuts inthe relative pay of the less-qualified are required,provided that taxpayers will accept the costs ofredistributing some consumption from themselves

16 Cuts in working time in order to share out available unskilled work would raise labour costs and threaten employment unlesspay were reduced commensurately, which this section of the labour force is least able to afford.

22


Dow

nloaded from


A Glyn

to those currently unemployed (Glyn and Ro wthorn,1994). Taxpayers benefit in terms of the public-sector infrastructure and services provided and interms of avoidance of the social divisions anddislocations which persistent mass unemploymentbring. The welfare state, moreover, through thebenefit system guarantees that the net cost to thegovernment of extra public expenditure which re-duces unemployment is less than the gross cost,considerably less where replacement ratios arehigh.17 However, some net cost is inevitable. Ifconsumption of the newly employed rises, butbudgetary considerations rule out deficits, thenthose currently employed must make a contributionin terms of reduced consumption.

The OECD regards this as infeasible, claiming,'New jobs must certainly be generated by the pri-vate sector, because in nearly all countries budgetdeficits and resistance to tax increases rule outsignificant expansion of the public sector' (OECD,1994a, p. 33). If they are correct about taxation,then it would not just be public-sector creation ofjobs that would be ruled out, but any measures

involving significant costs to the rest of society(such as subsidies for the creation of private-sectorjobs). This would imply that unemployment canonly be tackled by relying on market forces and thatthere is just no alternative to accepting the distribu-tional effects this generates, however harsh.

This pessimistic, not to say defeatist, conclusionparadoxically coincides with probably the biggestever transfer of resources to an economically dis ad-vantaged section of a country—the former GDR.The political conditions in this case are very spe-cial , but at least it illustrates that very large transferscan be acceptable.

There is no shortage of policies to tackle jobless-ness in general, and the plight of the least-qualified,inparticular.in ways which spread the costs. RichardFreeman's article in this issue shows the socialcatastrophe which can result from relying on mar-ket forces alone to solve contemporary joblessness.As Kalecki argued more than 50 years ago (1943),maintaining full employment is primarily a politi-cal problem.

REFERENCES

Abraham, K. (1991), 'Mismatch and Labour Mobility', in F. Padoa-Scioppa (ed.), Mismatch and Labour Mobility,Cambridge, Cambridge University Press.

Barrell, R. (ed.) (1994), The UK Labour Market, Cambridge, Cambridge University Press.— Caporale, G., and Sefton, J. (1994), 'Prospects for European Unemployment', in J. Michie and J. Grieve Smith

(eds), Unemployment in Europe, Academic Press.Baumol, W., Batey Blackman, S., and Wolff, E. (1989) Productivity and American Leadership, Cambridge, MA,

MIT Press.Bazen,S.,andMartin,J.(1991), "The Impact of the Minimum Wage on Earnings and Employment in France', OECD

Economic Studies no. 16, Paris, Organization for Economic Cooperation and Development.Bean, C. (1994) 'European Unemployment: A Retrospective', European Economic Review, 38, 523-34Bentolila, S., and Dolado J. (1994), 'Labour Flexibility and Wages: Lessons from Spain', Economic Policy,

18(April), 53-99.Berman, E., Bound, J., and Griliches, Z. (1993), 'Changes in the Demand for Skilled Labor within US Manufacturing

Industries: Evidence from the Annual Survey of Manufacturing', NBER Working Paper no. 4255, January.Blanchflower, D., and Oswald, A. (1994), 'Estimating the Wage Curve for the United Kingdom', Economic Journal,

September.Boltho, A., and Glyn, A. (1994), 'Macroeconomic Policies, Public Spending and Employment', Background Paper

for 1LO World Employment Report, Oxford, (mimeo).

17 The net cost to the government of public spending which employs additional workers is Gross Cost X (1 - Tax Rate) X (1 -Net Replacement Ratio). With an average tax rate (including social security contributions and indirect taxes) of one-third, andanetreplacementratio (post-tax benefits as aratio of post-tax wages) of one-half (alow figureby European standards—see OECD,1994fc, Tables 8.1 and 8.B.1) the net cost would be only one-third of the gross cost

23


Dow

nloaded from



Business Week(l994), 'Inequality: Howthe Gap between Rich and Poor is Hurting the American Economy', coverstory, 15 August, 77-83.

Calmfors, L. (1993) 'Lessons from the Macroeconomic Experience of Sweden', European Journal of PoliticalEconomy, 9.

— (1994), 'Centralisation ofWage Bargaining and Macroeconomic Performance—A Survey', OECD EconomicStudies, 21(Winter), 161-91.

— Driffill, J. (1988), 'Bargaining Structure, Corporatism and Macroeconomic Performance', Economic Policy,6.

Card, D., Katz, L., and Krueger, A. (1993), 'An Evaluation of Recent Evidence on the Employment Effects ofMinimum and Sub-minimum Wages', NBER Working Paper no. 4528.

Carlin, W., and Soskice, D. W. (1990), Macroeconomics and the Wage Bargain, Oxford, Oxford University Press.Commission of the European Communities (1993), Growth, Competitiveness and Unemployment, Brussels.Dreze, J., and Malinvaud, E. (1994), 'Growth and Employment: The Scope for a European Initiative', European

Economy.The Economist (1994), 'Inequality: For Richer, For Poorer', 5 November, 19-21.Elmeskov, J., and Pichelmann, K. (1994), interpreting Unemployment: The Role of Labour Force Participation',

OECD Economic Studies, 21, 139-60— MacFarlan, M. (1994), 'Unemployment Persistence', OECD Economic Studies, 21, 59-88.Freeman R., and Katz, L. (1994), 'Rising Wage Inequality: The United States vs. Other Countries', in R. Freeman

(ed.), Working Under Different Rules, New York, Russell Sage Foundation.Gittleman, M. (1994), 'Earnings in the 1980s: An Occupational Perspective'. Monthly Labour Review, July.Glyn, A. (1994), 'Does Profitability Really Matter', Oxford, (mimeo).— Rowthom,R.(1988), 'EuropeanUnemployment,CorporatismandStructuralChange',y4mer/cart£co/JOfffJc

Review, May.— — (1994),' European Employment Policies', in J. Michie and J. Grieve Smith (eds), Unemployment in Europe,

Academic Press.Green, F. (1994), 'Training, Inequality and Inefficiency', in A. Glyn and D. Miliband (eds), Paying for Inequality,

London, Rivers Oram.Gregg, P., and Machin, S. (1994), i s the UK Rise in Inequality Different?', in R. Barrell (ed.), The UK Labour Market,

Cambridge, Cambridge University Press.ILO (1995), World Employment Report, Geneva, International Labour Organization.Juhn, C , Murphy, K., and Topel, R. (1991), 'Why has the Natural Rate of Unemployment Increased Over Time',

Brookings Papers on Economic Activity, 75—126.Kalecki, M. (1943) 'Political Aspects of Full Employment', in J. Osiatynski (ed.) (1990), Collected Works ofMichal

Kalecki, vol. 1, Oxford, Oxford University Press.Layard, R., Nickell, S. J., and Jackman, R. (1991), Unemployment, Oxford, Oxford University Press.Machin, S. (1994), 'Changes in the Relative Demand for Skills in the Labour Market', University College, London,

(mimeo).Martins, J. (1993), 'Market Structure, International Trade and Relative Wages', OECD Economics Department,

Working Paper no. 134.Meidner, R. (1993), 'Why Did the Swedish Model Fail?', Socialist Register, London, Merlin Press.OECD (1989), Employment Outlook, Paris, Organization for Economic Cooperation and Development.— (1993a), Employment Outlook, Paris, Organization for Economic Cooperation and Development— (1993b), Education at a Glance, Paris, Organization for Economic Cooperation and Development— (1994a), The OECD Jobs Study: Facts, Analyses, Strategies, Paris, Organization for Economic Cooperation

and Development— (19946), The OECD Jobs Study: Evidence and Explanations (2 Vols), Paris, Organization for Economic

Cooperation and Development.— (1994c), Economic Outlook, December, Paris, Organization for Economic Cooperation and Development.— (1994a), Employment Outlook, Paris, Organization for Economic Cooperation and Development.Oxford Economic Papers (1994), special issue on Vocational Training, 46.Robinson, P. (1994), "The British Labour Market in Historical Perspective', CAP Discussion Paper 202, July.Rowthom, R. (1977), 'Conflict, Inflation and Money', Cambridge Journal of Economics, 1.— (1992), 'Corporatism and Labour Market Performance', in Pekkarinen, M. Pohjola, and R. Rowthom (eds),

Social Corporatism, Oxford, Oxford University Press.

24


Dow

nloaded from


A. Glyn

Schmitt, J., and Wadsworth, J. (1994), 'The Rise in Economic Inactivity', in A. Glyn and D. Miliband (eds), Payingfor Inequality, London, Rivers Oram.

Sachs, J., and Shatz, H. (1994), 'Trade and Jobs in US Manufacturing', Brookings Papers on Economic Activity, 1.Soskice, D. W. (1990), 'Wage Determination: The Changing Role of Institutions in Advanced Industrialized

Countries', Oxford Review of Economic Policy, 6(4).Summers, L. (1986),' Why is the Unemployment Rate So Very High Near Full Employment', Brookings Papers on

Economic Activity, 2.Summers, R. (1985), 'Services in the International Economy', in T. Inman (ed), Managing the Service Economy,

Cambridge, Cambridge University Press.Vartiainen, J. (1994), 'Can Nordic Social Corporatism Survive?', Stockholm, Trade Union Institute for Economic

Research, (mimeo).Wood, A. (1994), North-South Trade, Employment and Inequality, Oxford, Oxford University Press.

25


Dow

nloaded from


glyn (1995) - the assessment - unemployment and inequality

Documents