· Overqualification: reasons, measurement issues and typological affinity to unemployment Felix Büchel Extract from: Descy, Pascaline; Tessaring, Manfred (eds.). Training in Europe

Overqualification: reasons, measurement issues and typological affinity to unemploymentFelix Büchel

Extract from:

Descy, Pascaline; Tessaring, Manfred (eds.).

Training in EuropeSecond report on vocational training research in Europe 2000: background re-

port. (Cedefop Reference series). Luxembourg: Office for Official Publications ofthe European Communities, 2001 (3 volumes); ISBN 92-896-0034-0

Reproduction is authorised provided the source is acknowledged.

Additional information on Cedefop’s research reports can be found on:www.trainingvillage.gr/etv/research/index.asp

For your information:

❏ The background report to the second report on vocational training research in Europe containsoriginal contributions from researchers on different themes. It is published in English only. However,contributions were written in languages other than English, and readers interested in these shouldcontact Cedefop or the authors directly. A list of contents is on the next page.

❏ A synthesis report (about 450 pages) with additional research findings is being published in English,French, German and Spanish in the course of 2001.

Bibliographical reference of the English version:

Descy, Pascaline; Tessaring, Manfred. Training and learning for competence. Second report on vo-cational training research in Europe: synthesis report. Cedefop Reference series. Luxembourg: Of-fice for Official Publications of the European Communities, 2001. ISBN 92-896-0029-2

❏ In addition, an executive summary in all EU languages (about 50 pages) is available from Cedefopfree of charge.

The background and synthesis reports are available from national EU sales offices or from Cedefop(www.trainingvillage.gr/etv/publication/form1.asp ).

Prices in Luxembourg (excluding VAT): EUR 21 for the background report (3 volumes); EUR 19 for thesynthesis report.

Price for the synthesis report and the background report in Luxembourg (excluding VAT) in a box: EUR29.50.

Addresses of the national sales offices can be found at: http://eur-op.eu.int/general/en/s-ad.htm.

For further information contact:

Cedefop, PO Box 22427, GR-55102 ThessalonikiTel.: (30-31) 490 111Fax: (30-31) 490 102E-mail: [email protected] , Homepage: www.cedefop.eu.int ,Interactive website: www.trainingvillage.gr

Contributions to the background report of the second research report

VOLUME 1

Part One: VET systems, coordination with the labour marketand steering

Steering, networking, and profiles of professionals in vocationaleducation and training (VET)Lorenz Lassnigg

Financing vocational education and trainingAndy Green, Ann Hodgson, Akiko Sakamoto, Ken Spours

How to improve the standing of vocational compared to generaleducation. A collaborative investigation of strategies and qualifica-tions across EuropeJohanna Lasonen, Sabine Manning

Certification and legibility of competenceAnnie Bouder, Laurence Coutrot, Édith Kirsch, Jean-Louis Kirsch,Josiane Paddeu, Alain Savoyant, Emmanuel Sulzer

The changing institutional and political role of non-formal learning:European trendsJens Bjørnåvold

The problems raised by the changing role of trainers in a Euro-pean contextMara Brugia, Anne de Blignières

Part Two: Lifelong learning and competences:challenges and reforms

Lifelong learning - How the paradigm has changed in the 1990s.Martina Ní Cheallaigh

Training for new jobs: contents and pilot projectsJeroen Onstenk

Vocational training and innovative practices in the environmentalsector. A comparison of five EU Member States, with specimencasesRoland Loos

Company-based learning in the context of new forms of learningand differentiated training pathsPeter Dehnbostel, Gisela Dybowski

VOLUME 2

Part Three: Training and employment in a company perspec-tive

Globalisation, division of labour and training needs from a com-pany viewJohan Dejonckheere, Geert Van Hootegem

Training, mobility and regulation of the wage relationship: specificand transversal formsSaïd Hanchane (with the assistance of Philippe Méhaut)

The employment and training practices of SMEs. Examination ofresearch in five EU Member StatesPhilippe Trouvé et al.

Human resource development in Europe - At the crossroadsBarry Nyhan

Reporting on human capital: objectives and trendsSven-Åge Westphalen

Vocational training research on the basis of enterprise surveys: aninternational perspectiveLutz Bellmann

Part Four: Employment, economic performance and skillmismatch

The skills market: dynamics and regulationJordi Planas, Jean-François Giret, Guillem Sala, Jean Vincens

Economic performance of education and training: costs and bene-fitsAlan Barrett

Unemployment and skills from a dynamic perspectiveJoost Bollens

Overqualification: reasons, measurement issues and typologicalaffinity to unemploymentFelix Büchel

Forecasting skill requirements at national and company levelsRobert A. Wilson

VOLUME 3

Part Five: Individual performance, transition to active life andsocial exclusion

Training and individual performance: evidence from micro-econometric studiesFriedhelm Pfeiffer

The effect of national institutional differences on education/trainingto work transitions in Europe: a comparative research project(Catewe) under the TSER programmeDamian F. Hannan et al.

Education and labour market change: The dynamics of educationto work transitions in Europe. A review of the TSER programmeDamian F. Hannan, Patrick Werquin

Selection, social exclusion and training offers for target groupsJan Vranken, Mieke Frans

Training and employment perspectives for lower qualified peopleJittie Brandsma

Part Six: VET research activities outside the European Union

Research on vocational education and training at the crossroadsof transition in central and eastern EuropeOlga Strietska-Ilina

VET research in other European and non-European countriesUwe Lauterbach et al.

Annex: VET related research on behalf of theEuropean Commission

Research on vocational education and training in the current re-search framework of the European CommissionLieve Van den Brande

Synopsis of selected VET-related projects undertaken in theframework of the Leonardo da Vinci I programmeCedefop

Targeted socio-economic research (TSER): project synopsesCedefop

Overqualification: reasons, measurement issues

453

Overqualification: reasons, measurement issues and typo-logical affinity to unemployment

Felix BüchelAbstractThis paper focuses primarily on two essential aspects of overqualification research: the reasonswhy people find themselves in underskilled jobs and the problems involved in establishing avalid means of identifying such a situation.

Starting with an extensive terminological discussion, the paper briefly introduces the varioustheories explaining the phenomenon of overqualification and discusses some influential fac-tors that have rarely been examined in any work of literature, namely institutional andmacroeconomic conditions and labour utility and productivity assessments on the supply anddemand sides.

The next step is a systematic examination of the problems involved in identifying underskilledjobs. It begins with an explanation of the two main measurement strategies (the ‘objective’ andthe ‘subjective’ approaches) and their diverse variants including unorthodox measurement proc-esses. Thereafter, an extension of the subjective measurement strategy is introduced; in terms ofvalidity, it is probably superior to the forms of categorisation that have been used hitherto.

The paper goes on to provide a comprehensive review of research literature, beginning with theorigins of the debate in the educational expansion of the 1970s and 1980s. It retraces the aca-demic discussion on the overqualification problem in the United States, then portrays the Ger-man situation as an example of the early development of overqualification research outside theUnited States. It goes on to introduce more recent work from the United States and Europe,including international comparative studies.

On the basis of longitudinal data from Germany, the theoretical analogy between unemploy-ment and overqualification is empirically tested. The first part of this test focuses on foursubjective indicators, which are tested for divergences between unemployed people, people inunderskilled jobs and people whose jobs match their qualifications; the four indicators, whichare already established in the field of static unemployment research, are problem-solving skills,morale, concerns about the future and participation in the political process. In the second partof the test, the set of instruments used in dynamic unemployment research is applied to analyseoverqualification. The analysis is based on examples of movements from various types of em-ployment status into underskilled work and from underskilled work into unemployment (down-ward career moves) as well as transitions from underskilled to adequately skilled jobs (up-ward career moves). Particular interest attaches to the question of the extent to whichunderskilled work and unemployment are alternative situations. Finally, the paper examineslonger-term income effects of overqualification. Here, too, the main question is whether a pe-riod spent in an underskilled job creates income effects which – as should theoretically be thecase – lie somewhere between the effects of a period spent in an adequately skilled job and thoseof a spell of unemployment.

The study ends with a catalogue of methodological conclusions, aimed at employment research-ers, and substantive considerations, aimed at policymakers in the fields of education and em-ployment.

Felix Büchel

454

Table of contents

1. Introduction ..................................................................................................................... 456

2. Subject of the study and definitions .......................................................................... 459

3. Theoretical reflections on the causes of overqualification .................................. 4613.1 Overqualification from the neoclassical perspective ........................................................ 4613.2 Decision-making in the event of an individual mismatch between a person’s skill

profile and the job description ........................................................................................... 461The employee’s decision ..................................................................................................... 462The employer’s decision ..................................................................................................... 462

3.3 Explanatory methods based on partial analysis .............................................................. 4633.3.1 Established methods ................................................................................................. 463

The human capital theory ........................................................................................ 463The assignment theory ............................................................................................. 463Filtering, signalling and screening approaches ...................................................... 464Job-competition model .............................................................................................. 465The job-matching theory ........................................................................................... 467Career mobility theory .............................................................................................. 468The theory of differential overqualification............................................................. 470The job-search theory ............................................................................................... 473

3.3.2 Unorthodox approaches ............................................................................................ 473Segmentation approach ............................................................................................ 473The Marxist dialectical approach............................................................................. 475

3.4 Extraneous conditions ........................................................................................................ 4763.4.1 Institutional regulation ............................................................................................ 4763.4.2 The production structure and conditions in the labour market ............................. 4783.4.3 Employees’ utility preferences ................................................................................. 4783.4.4 Employers’ productivity considerations................................................................... 479

4. Strategies for measuring the educational adequacy of a job .............................. 4824.1 Existing measurement strategies...................................................................................... 482

4.1.1 The ‘objective’ approach ............................................................................................ 4824.1.2 The ‘subjective’ approach .......................................................................................... 4884.1.3 Unorthodox measurement strategies ...................................................................... 489

4.2 An innovative approach ..................................................................................................... 491

5. Overqualification: a literature survey ...................................................................... 4915.1 Origins of the discussion .................................................................................................... 491

5.1.1 The U.S. discussion from the sixties to the eighties ............................................... 4915.1.2 The early discussion outside the United States: the case of Germany .................. 493

5.2 Present state of research ................................................................................................... 4955.2.1 Recent studies relating to the United States and Canada ..................................... 4955.2.2 Recent studies relating to the countries of the European Union ........................... 496

The Netherlands ....................................................................................................... 496Germany .................................................................................................................... 498Other countries of the European Union .................................................................. 501

5.2.3 International comparative studies ........................................................................... 503


455

6. Overqualification and unemployment: The case of Germany ............................. 5046.1 Research approach ............................................................................................................. 504

6.1.1 Database .................................................................................................................... 504Information on the education and training of foreigners ....................................... 504General case selection............................................................................................... 505

6.1.2 Categorisation of overqualification .......................................................................... 505Identifying underqualification ................................................................................. 507The degree of overqualification ................................................................................ 507Validity checking ....................................................................................................... 508

6.1.3 The evaluation process ............................................................................................. 5086.2 Subjective indicators: differences between adequate employment,

overqualification and unemployment................................................................................ 5096.2.1 Problem-solving capacity .......................................................................................... 5096.2.2 Morale ........................................................................................................................ 5096.2.3 People’s concern about their own economic prospects ............................................ 5106.2.4 Political involvement ................................................................................................ 510

6.3 Transition to inadequate employment .............................................................................. 5106.3.1 Preliminary remarks on the dynamics of individual overqualification ................. 5106.3.2 Descriptive findings .................................................................................................. 5116.3.3 Multivariate findings ................................................................................................ 512

6.4 Selected transitions from overqualification ...................................................................... 5136.4.1 Transition to unemployment .................................................................................... 514

Transition to unemployment from inadequate or adequate employment ............. 514Transition from adequate employment to unemployment –directly or via overqualification ............................................................................... 514

6.4.2 Transition to adequate employment ........................................................................ 515Transition to adequate employment from overqualification or unemployment .... 515Transition from unemployment to adequate employment –directly or via overqualification ............................................................................... 515

6.5 Income effects of overqualification .................................................................................... 5166.5.1 Income growth among people who are initially overqualified................................ 5166.5.2 Erosion of human capital through overqualification .............................................. 516

7. Conclusions ...................................................................................................................... 5177.1 Methodological consequences for employment researchers ............................................. 5177.2 Implications for education and employment policies ....................................................... 519

7.2.1 Implications for education policy ............................................................................. 519Dual vocational training ........................................................................................... 520Higher education ....................................................................................................... 521

7.2.2 Implications for employment policy ......................................................................... 522

Summary .................................................................................................................................... 524

Bibliography .............................................................................................................................. 528

Annex .......................................................................................................................................... 544Classification model of overqualification ................................................................................... 544Empirical results for Germany ................................................................................................... 545

Felix Büchel

456

1. Introduction2

It has gradually come to be recognised and isnow an unchallenged truism that the humancapital a society has accumulated in the formof education, and especially vocational edu-cation, is a crucial location factor in the con-text of global economic competition. If thematter is considered in this light, however,the entire human capital created through theeducation system is not what counts but onlythat which is ‘productively’ invested in theeconomy. Because of the mass unemploymentthat prevails in most European countries,substantial volumes of human capital are ly-ing dormant. The result of this is economicunderachievement by the Member States ofthe European Union. The ‘superfluous’ skillsand qualifications, in other words those forwhich there is no demand in the labour mar-ket, are not just temporarily sidelined but arealso rapidly devalued as a result of disuse andthe consequent absence of training opportu-nities. This hysteresis phenomenon, in whichthe unbalanced labour market is both an ef-fect of past unemployment and a cause of fu-ture unemployment, causes the problems toescalate with the passage of time.

The losses to national economies and the so-cial problems that are generated by unem-ployment are clearly recognised, not only inemployment research but by the general pub-lic too. One of the main aims of this study isto draw attention to the fact that the actualsurplus of vocational skills and qualificationsproduced by the education system consider-ably exceeds the surplus indicated by unem-ployment statistics, even if analyses of hid-den labour reserves are taken into account.

Unused and therefore unproductive skills alsocome into the reckoning when individualshave jobs for which they are palpablyoverqualified. The analogy between suchunderskilled work, where the employee’s levelof training exceeds the job requirements, andunemployment is unmistakable: the entirehuman capital of jobseekers lies idle, whileonly part of the human capital of overqualifiedemployees lies idle. The latter phenomenonis also absolutely relevant to national econo-mies; its significance can be measured if thenumber of overqualified employees3 is multi-plied by the amount of skill wastage. The skillwastage may be empirically calculated, forexample, as the difference between the remu-neration of overqualified employees and theamount they would be earning if their jobmatched their skill and qualification level(see, for example, Duncan and Hoffmann,1981; Daly et al. (forthcoming) and Büchel andWeißhuhn, 1997c and 1998). It is thereforeno coincidence that the editors of the presentreport have combined the sections on unem-ployment and overqualification into a singlesection. A structural analogy between unem-ployed status and overqualification can alsobe derived from the terminological definitionlaid down by the OECD,4 which uses ‘unem-ployment and underemployment’ as a concep-tual entity within its system. It divides the

‘There can never be too much education’ (Rumberger, 1981b, p. 7).‘He is an overeducated s.o.b.’ (Harry S. Truman on J. William Fulbright)1

1 s.o.b. = son of a bitch. Source: Die Zeit of 17 Feb-ruary 1995, p. 9.

2 The author thanks Manfred Tessaring for mostimportant observations and Johannes Giesecke forhis valuable assistance in implementing an en-hanced version of literature management software.

3 The overqualified percentage of the labour forcevaries widely, depending on the measurement proc-ess (on this central problem of overqualificationresearch, see for example Smith (1986), Groot andMaassen van den Brink (forthcoming) and section5 of the present study, which goes into some detailon the subject). The crucial point, however, is thatthis percentage is so high in all the Western in-dustrialised countries that it cannot be ignored.For example, it exceeds the national unemploy-ment rate in every country. Hartog (1997) andBorghans et al. (1998), as well as Groot andMaassen van den Brink (forthcoming) and Hartog(1999b) provide international comparisons of thefrequency with which overqualification is observed.

4 Resolution I of the International Conference ofLabour Statisticians (ICLS) on the ILO: see OECD,1995a, pp. 44-45.


457

category ‘underemployment’, into two sub-categories, namely ‘visible’ and ‘invisible’ un-deremployment. The first term relates to peo-ple who are employed but who have fewerworking hours than they would wish (see forexample Terry, 1981); the second term de-scribes underskilled work (‘... refers to indi-viduals who are working in jobs where theirskills are not adequately utilised’).5

At first sight, both of these subcategories ap-pear almost too disparate to be two halves ofthe same whole. Visible underemployment isstructurally closer to unemployed status,since the individual is working less than heor she would wish; in structural terms, it isakin to the German status of Kurzarbeit, ap-plicable to employees on short time. On closerinspection, however, typological analogies areimpossible to overlook. In neither of the situ-ations can the full economic value of an indi-vidual’s vocational skill be realised; in the caseof visible underemployment, this is becauseof an unwanted part-time working arrange-ment, which is ultimately due to restricteddemand for the available skill profile; in thecase of invisible underemployment, the rea-son is the low skill level required for the jobsthat overqualified individuals take, presum-ably for want of something better.

The analogy between overqualification in gen-eral and unemployment lies in the fact thatboth are due to the aforementioned lack ofdemand for particular skills – with unemploy-ment reflecting a total absence of demand andoverqualification a shortfall in the volume ofdemand. The deficit in market demand forvocational skills and qualifications which isa consequence of chronic mass unemployment

and which numerous studies have amply dem-onstrated is therefore systematically under-estimated whenever overqualification is leftout of the equation.

This typological analogy is presented in tabu-lar form in some of the more recent publica-tions of the U.S. Bureau of Labor Statistics.Hecker (1992, Table1, p.4), for example,presents annual figures under the followingthree headings: ‘All graduates in the laborforce’, ‘Graduates in jobs that require a de-gree’, ‘Graduates in jobs that do not require adegree or who are unemployed’, the last col-umn being divided into two subcolumns forgraduates in non-graduate posts and unem-ployed graduates.

The structural analogy between unemploy-ment and overqualification, however, hasscarcely featured in employment research andits empirical surveying strategies. When re-searchers examine underskilled work andunemployment simultaneously or accord asimilar level of priority to both, they are in-clined to assume rather intuitively that bothphenomena are serious problems for thosewho are affected by them and that their wide-spread occurrence on a macroeconomic scalemay be regarded as a symptom of a profounddisorder in the labour market.6 One exceptionis an earlier study of mine (Büchel (1998b),which undertakes the first systematic exami-nation with the instruments of dynamic un-employment research based on career cycles,assessing for example the probability of ac-cess to, continuation in and withdrawal fromemployment as well as the long-term effectsof a period of overqualification.

The question arises as to why the causal linkbetween dormant vocational skills and lowerliving standards has been clearly recognisedand adequately researched in connection withunemployment, whereas considerably lessattention has been devoted to the problem ofinsufficient coordination between the educa-

5 It should be noted that the term ‘underemploy-ment’ is sometimes used in a different sense. Ruiz-Quintanilla and Claes (1996), for example, includepart-time work, temporary work and unemploy-ment in this category, ignoring the fact that thefirst two forms, especially part-time work, are of-ten undertaken on a voluntary basis; this broaddefinition, however, is contrary to the intention un-derlying the OECD terminology. Another entirelyunorthodox version is proposed by Lichter (1988),who uses a category labelled ‘underemploymentby low income’.

6 See for example Bielinski et al. (1994 and 1995)in relation to eastern Germany, which seems toillustrate this phenomenon particularly well, orCedefop (Tessaring, 1998) in the European context.

Felix Büchel

458

tion system and the labour market, a prob-lem that manifests itself in the form ofoverqualification.7

There are probably three main reasons. Firstof all, unemployment is a particularly flagrantwaste of human capital, because the loss isabsolute. Secondly, unemployment imposes aburden on national economies not only be-cause the unemployed do not contribute to acountry’s productive output but also becausepart of the national product has to be divertedinto a fund for the payment of unemploymentbenefits. And thirdly – an aspect that prob-ably should not be underestimated – the pre-cise volume of dormant human capital thatresults from unemployment has been statis-tically measured and recorded. Official sta-tistics relating to overqualification, on theother hand, do not flag any problem, especiallynot in the highly aggregated form in whichsuch statistics are characteristically producedand studied.8 Take the example of a man witha doctorate in philosophy who works as a taxidriver. He is not unemployed and thereforereceives no state benefits – on the contrary,as far as the skill-related labour-market sta-tistics are concerned, he even helps to furnishevidence that the economy employs philoso-phers and hence that there is a demand forphilosophers. This signal is then transmittedto universities and their arts faculties andphilosophy departments.

If dormant skills in the labour market aretaken into consideration, it follows from thepreceding observations that a new dividingline has to be drawn when we examine thelabour market. The conventional boundarybetween jobseekers and employed persons

must give way to a division between job-seekers plus people in underskilled jobs onthe one hand and employed persons whosejobs match their skill and qualification levelson the other.9 It is evident that this type ofapproach must take account of the fact that aconsiderably higher volume of skill lies dor-mant when individuals are unemployed thanwhen they are in overqualification, nor can itbe denied that unemployment has a fargreater impact on both the individual andsociety than the phenomenon of overquali-fication. Nevertheless, as far as the utility ofacquired vocational knowledge and skill isconcerned, there is still a typological analogybetween unemployment and overqualifi-cation.

The present study pursues three principalobjectives. Once the terminological conceptshave been clarified, we shall present the firstcomprehensive investigation designed to es-tablish which of the competing labour-mar-ket theories can help to explain the persist-ence of overqualification (section 3). Apolitically guided improvement of the inter-action between the education system and thelabour market, and hence an increase in theefficiency of the education system, cannot beachieved without knowledge of these causal

7 The contrast here between the public interestand the present state of research is particularlystriking (cf. Büchel, 1996b).

8 Although it is theoretically possible to differen-tiate, i.e. to break official data down into narrowercategories, it is often impossible in practice, be-cause such differentiation presupposes access tothe original figures from which the statistics werecompiled, and access to these figures is frequentlyrestricted by the authorities. For an example ofthis type of approach in connection with the sub-ject of the present study, see Plicht et al., 1994.

9 On this point, compare the generally analogousbut considerably more radical and empirically al-most unworkable approach presented by Müllerand Beck (1993): ‘We are speaking [...] of under-utilisation of the production factor labour whenwe refer to members of a country’s active popula-tion who either do not have a job, even though theyare prepared to work for the going rate of pay (un-employment type A) or do have a job but are notused efficiently in the production process (this iswhat we call hidden unemployment; unemploy-ment type B)’ (pp.54-55, including text in paren-theses). The employees whom Müller and Beckdefine as ‘not used efficiently’ come from a widevariety of categories, such as employees who aresurplus to requirements but cannot be dismissed,employees whose job dissatisfaction is reflected inreduced productivity and ‘people who, in accord-ance with the accepted norms of administrativetheory, remain in posts in which their services areeffectively superfluous’ [!] (p.54). For a critique ofthe conventional dichotomous division of the ac-tive population into employed and unemployed per-sons in a broader context, see also Sulliven, 1981,pp.335ff.


459

links.10 To supplement this investigation, weshall also review the current state of empiri-cal research (section 4).

Our second objective is to discuss in detail themeasurement problems associated with anyattempt to define overqualification. It willemerge from this discussion that the volumeof overqualification is less accurately meas-urable than unemployment. At the same time,it will become apparent that these measure-ment problems not only make internationalcomparisons in the field of overqualificationmore difficult than those relating to unem-ployment; the diversity of measurement strat-egies, to which varying degrees of validity canbe ascribed, also leads us to the unsatisfac-tory conclusion that there is scarcely a con-sensus on the volume of overqualificationwithin the various national economies. Thispart of the study concludes with the presen-tation of our own model for the categorisa-tion of overqualification.

The third objective of the study is to substan-tiate empirically the postulated analogy be-tween jobseeker status and overqualifiedemployee status. To this end, section 6 pur-sues an integrated approach, contrasting sub-jective indicators that are firmly establishedin employment research, such as satisfactionwith life and concerns about the future, forunemployed people, people in underskilledjobs and those whose jobs match their quali-fication levels. The section also contains a lon-gitudinal analysis of movements betweenthese three types of employment status witha view to establishing whether significantexchange processes are observable, particu-larly between unemployment and overquali-fication, and whether the gap between thesetwo situations is narrower than that between

skill-matched employment on the one handand unemployment or overqualification on theother. Section 7 draws conclusions from thepreceding reflections and findings.

2. Subject of the studyand definitions

The term ‘overqualification’ is generally usedto describe a situation in which the knowl-edge and skills acquired through the educa-tion system are not exploited to the full. Thisdefinition is clearly very vague. For that rea-son, German researchers at any rate do notspeak of overqualification unless the requiredskill level for the job is so far below the for-mal qualification of the employee that thework could obviously be performed by some-one with a lower level of formal qualification.

This interpretation is not unconnected to thedegree of formalisation of vocational training.The more institutionally standardised thequalification, the more plausible this ap-proach appears. Applied to a rigidly seg-mented system such as the German one,which is divided into vocational training (non-academic) and higher vocational education(academic), this means that, on the one hand,people who have undergone vocational train-ing – in the form of an apprenticeship, for in-stance – and are now in jobs for which no for-mal vocational qualification is required (e.g.a carpenter who works as a car-park attend-ant) and, on the other hand, graduates in jobswhich could be done by people without highereducational qualifications (e.g. an engineer-ing graduate working as a technician or evenas a fast-food chef) are considered to be inoverqualification.

As the definition shows, the important factoris the vertical dimension of occupational flex-ibility, not the horizontal dimension.11 Al-though the latter also results in the disuse ofknowledge and skills acquired during thetraining process (if a qualified nurse, for ex-ample, works as a medical technician), the

10 Despite a comprehensive body of relevant Ameri-can literature, this type of in-depth analysis hasyet to be performed in the United States. As Heckersays, ‘Currently, we do not know how many col-lege graduates take noncollege-level jobs becauseof labor market supply-and-demand conditions,how many do so because their individual educa-tional backgrounds limit their options, and howmany do so through deliberate choice.’ (Hecker,1995b, p.41).

11 On this distinction, see also Plicht et al., 1994,p.178).

Felix Büchel

460

type of ‘skill spillage’ that occurs when thetraining curriculum is out of alignment withthe job description is considered to be inher-ently unavoidable when jobs are allocated tojobseekers.

It may be assumed that this form of horizon-tal mismatch will become increasingly com-monplace as the industrial division of labourgrows in complexity. Accordingly, a high inci-dence of horizontal mismatches in a nationaleconomy can scarcely be cited as evidence ofan inefficient education system. In Germany,for example, a very high percentage of theactive population state that they are not work-ing in the occupation for which they weretrained. Not least among the reasons for thisis the fact that the dual system of alternatingpractical and theoretical training is based ona highly refined catalogue of occupational dis-ciplines; an employee can therefore claim notto be doing the job for which he was trainedeven though the content of his actual job isvery similar to the one in which he is quali-fied. Such ‘skill spills’, i.e. elements in thetraining curriculum that are not needed atwork, also tend to affect every person whostarts a new job.

So the decisive question is whether an em-ployee’s post could also have been filled bysomeone with a lower level of formal qualifi-cation, which would mean that the employ-ee’s vocational training represents a wastedpersonal investment, at least in part. In thecase of the qualified nurse who was mentionedabove, however, this situation is unlikely toobtain, since it may be assumed that shewould not have obtained her post as a medi-cal technician if she had not been trained toan equivalent standard in an associated dis-cipline; her fellow technicians may thereforebe expected to possess a similar level of quali-fication.

In the literature from the English-speakingworld in which the terminology of employ-ment research is coined, the terms ‘over-education’ and, to a lesser extent, ‘over-qualification’ have won acceptance as labelsfor the phenomenon described in this study.It should be noted, however, that in the ini-tial phase of this discussion (see point 5.2.1

below) and even thereafter, in a few cases (seefor example Dooley (1986) or Lambropoulosand Psacharopoulos (1992), who implicitlysaid the same thing, albeit in another con-text and without using the term ‘over-education’) these concepts were used not onlyto refer to an individual in an underskilledjob but also to refer in macroeconomic termsto a surfeit, or alleged surfeit, of paper quali-fications in the labour market, which mani-fests itself, for instance, in diminishingreturns from higher qualifications (cf. Rum-berger, 1981c, p.294). However, other termsin frequent use, namely ‘skill underutilization’(see for example Kalleberg and Sørensen,1973, p.217, as well as Staines and Quinn,1979, p. 8) and ‘surplus schooling’ (see forexample B.Tsang et al., 1991), relate explic-itly to the individual. The term ‘mismatch’occurs occasionally; from each of the contextsin which it is used, it is clear that the mis-match is between the formal qualificationlevel and the skill level of the job (see for ex-ample Clogg and Sullivan, 1983, p.121); nowand then the term ‘occupational mismatch’ isalso encountered (Clogg, 1979, and Sullivan,1978). This concept, however, seems to posesemantic problems, because in the strictly lit-eral sense it includes overskilled jobs,12 whichare irrelevant in the context of our study andwhich, in a different framework, would haveto be regarded as overqualification.13

Other designations have not stood the test oftime, such as ‘overtraining’ (Kalleberg andSørensen, 1973) or the term ‘nominal over-education’, which Halaby used (1984, p.48) andascribed to Clogg and Shockey (1984), although

12 Such a situation would arise, for example, if theholder of a master craftsman’s certificate were theonly non-graduate member of the managementteam of a sizeable construction company.

13 The difference between the theoretical ap-proaches to the phenomena of underskilled andoverskilled work is reflected, for example, in thefact that the proponents of the theory of careermobility (Sicherman and Galor, 1990) explicitlystate that they can only explain the persistence ofoverqualification, not the occurrence of un-derqualified activity (cf. Sicherman, 1991, pp.109-110).


461

this reference could not be traced. The samefate awaited the terms Grauzonentätigkeit(‘grey-area activity’), which Schlegelmilch usedin 1982, and ‘inappropriateness of skills’(Tessaring, 1998). Nor could Mincer win accept-ance for his proposed alternatives to ‘over-education’. He criticised the tendency to con-fuse ‘education’ with ‘schooling’, which wasonly part of an individual’s education, and sug-gested the terms ‘overschooling’ and ‘mis-schooling’ (Mincer, 1984, p.208).

Finally, there is surely a touch of irony in thefact that the expression ‘invisible underem-ployment’ which was quoted from the OECDterminology in the introduction to this studyis conspicuous by its virtual absence from thebody of literature on the subject of over-qualification – nomen est omen!

3. Theoretical reflections onthe causes of overqualification

3.1 Overqualification from theneoclassical perspective

From a neoclassical point of view, the phe-nomenon of overqualification, like that ofunemployment, is a ‘non-event’. Whereoverqualification occurs, the neoclassical viewis that it can only be the symptom of a tempo-rary imbalance. This opinion is expressed bythe authors of earlier conventional macro-economic studies on overqualification (see forexample Freeman, 1976b).

In the longer-term adjustment process, forwhich, these authors believed, political sup-port would be needed, the stiffer competitionfor adequately skilled jobs would lead to payreductions at the top end of the labour mar-ket. This in turn would induce companies toadapt their production structures, the generaleffect of which would be an increase in de-mand for higher qualifications. On the sup-ply side – always assuming perfect informa-tion – the diminishing return on higher levelsof skill and qualifications would make ad-vanced training a less attractive proposition.The demand-side adjustment would thus bereinforced by a reduction in the supply ofskilled labour.

As always happens in neoclassical models, astable balance would be restored; any remain-ing imbalances between the supply of skilledlabour and the demand for skilled labour,which would manifest themselves in unem-ployment and underskilled work, would thenbe no more than frictional loss.

Criticism of this view begins at the samepoints as the familiar criticism that has beenlevelled at the neoclassical understanding ofunemployment. Market operators seem evenless likely to conform to neoclassical theoryin their responses to overqualification thanthey do in responding to unemployment. Theinformation gaps are even wider, which meansthat longer reaction times may be expectedon both sides of the market. Trusting solelyin the healing powers of market forces to over-come overqualification ‘in the long term’therefore seems just as inadvisable as it is inthe case of unemployment; Keynes’ criticismof the neoclassical insistence on patience isvery apt here too. As he said, ‘In the long runwe are all dead’.

3.2 Decision-making in the event of anindividual mismatch between aperson’s skill profile and the jobdescription

In the general empirical literature on the sub-ject of overqualification, the labour-markettheories referred to in point 3.3.1 below aregenerally cited as possible ways of explain-ing the phenomenon. These various ap-proaches are then evaluated in the light ofresearch findings to establish their empiricallegitimacy. The observable degree of concur-rence among all the relevant empirical stud-ies is remarkably high.

None of the studies presented in section 4 fo-cuses on the precept that the interests of bothemployers and employees must always be rec-onciled in any job-matching process, a pre-cept which is very important if we are to graspthe overall picture (Franz, 1991, chapter 6).Most of the attempts to explain the persist-ence of overqualification that are enumeratedin point 3.3.1 below are only able to explainthe calculation underlying the decision takenby one side; apart from the theory of career

Felix Büchel

462

mobility – and even in that theory the inter-ests of the employers’ side are only implicitlyassumed – there is not a single theory thatcan explain why the same decision is takensimultaneously by different players.

Equally scant attention is devoted in the bodyof empirical literature to investigation of thedecision-making options of both sides in thelabour market. In fact, the nature of the deci-sion-making problem differs quite markedlybetween the supply and the demand side.

The employee’s decision

For the employees’ side, the implicit alter-native to underskilled work is normally un-employment or withdrawal into the hiddenreserve (for empirical evidence see Schlegel-milch, 1982 and 1983b). According to thispremise, the main difference betweenoverqualified employees and the unemployedis the former’s unconditional desire for a job,what Schlegelmilch refers to as an ‘I’d havedone almost anything’ attitude (Schlegel-milch, 1982, p.414).

It must clearly be assumed – implicitly, onceagain – that a person’s acceptance of anunderskilled job will have been preceded by afruitless search for more suitable employment.This perspective, however, systematically ob-scures two other possible scenarios. Firstly, aspontaneous switch from another form of eco-nomic inactivity (that of the non-jobseeker) tooverqualification is conceivable if an accept-able underskilled job offer were to materialise– an unexpected opportunity, obviously – andif that job seemed likely to benefit the indi-vidual more than continued inactivity.

It is also conceivable that underskilled workcould be a genuine alternative to adequatelyskilled employment. This implies a ‘voluntary’choice of lower job status. Teichler (1994,p.30), in a résumé of the findings of a studyby Teichler et al. (1992), surprisingly reportsthat the majority of the academic respondentswho were in underskilled posts had voluntar-ily chosen that employment status; Hecker(1995b, p.41) also discusses the same optionin the U.S. context, though without empiri-cally testing a hypothesis.

Such a decision could be taken, for example,if advantageous non-monetary working con-ditions, such as a lighter workload, a shorterjourney from home to work, a job with highregional status, shorter working hours (if pre-ferred), etc., outweighed the monetary disad-vantages of an underskilled job in relation toadequately skilled employment and if incomemaximisation were not the basis of thejobseeker’s decision (see point 3.4.3 below,where this factor is addressed in greater de-tail). Moreover, it might also happen, albeitmore rarely, that a higher net income can beearned – at least in the short term – fromoverqualification than could be obtained froma well-matched job (cf. Lucas, 1977).

Although all the indicators suggest that thisfavourable pay differential will not be sustain-able in the long run (see for example Büchel,1994b), if the jobseeker does not possess thisinformation or only wants to do the job for alimited time, it is entirely logical for him orher to accept an underskilled job as an alter-native to more appropriate employment. Thereis empirical evidence of both the aforemen-tioned types of decision in the present report.

The employer’s decision

The motives behind the employer’s decisionalmost never feature as a subject of empiricalliterature either. Nevertheless, this decision isquite different to that of the employee. Themain alternative in empirical terms to appoint-ing an overqualified candidate is most prob-ably the selection of an applicant whose skillsand qualifications are commensurate with thejob description. This, however, presupposes theabsence of restrictions on the supply side.

It is entirely conceivable, for example, thatthe local labour market might be unable todeliver a ‘suitable’ person for a particular job,i.e. one whose skills and qualifications matchthe job profile, perhaps because of the spe-cific nature of the job in question. At the sametime, the job might not be important enoughto warrant the additional cost of casting therecruitment net beyond the local area. In thissituation it would be futile to investigate theemployer’s motives for choosing an overquali-fied applicant.


463

This turns the spotlight onto another optionthat is open to employers: if there is not evenan overqualified candidate for a post, or ifoverqualification is automatically regarded asa reason for rejection of an application, theemployer may simply decide not to fill the postat all.14

When the empirical effectiveness of the com-peting attempts to explain the persistence ofoverqualification is assessed, due considera-tion must be given to the difference betweenthe parameters governing the employer’s andthe employee’s decisions. This means that,even if an explanatory method (based on oneparty’s decision) proves to be especially effec-tive, it is still only a partial analysis, and thepersistence of the phenomenon will not befully explained unless plausible explanationsare found for the behaviour of the other party.

3.3 Explanatory methods based onpartial analysis

3.3.1 Established methods

The explanatory methods we call ‘established’here are those that have featured in empiricalstudies in the field of overqualification re-search. This, however, does not automaticallyimply anything about the popularity of a givenmethod or its empirical substantiation.

The human capital theory

The human capital of employees comprisesnot only the certified knowledge they haveacquired at schools or technical colleges but

also contains other components, which arecreated as the employees gather work expe-rience or which reflect their length of servicewith a company, for example. If these diversecomponents are interchangeable, it is conceiv-able that various people whose componentsare mixed in different proportions might pos-sess an equal stock of human capital. In sucha case, measuring the extent of a mismatchon the sole basis of formal qualifications wouldnot furnish proof of inefficient skill deploy-ment in the labour market.

This more methodological aspect of over-qualification, however, only comes into play ifa DOT/GED approach is adopted to measurethe degree of overqualification; such an ap-proach is not based on employees’ subjectiveself-reported estimates but on a standard setof measures of required schooling for variousoccupational categories (see subsection 4.1below). Accordingly, the human capital ap-proach (Becker, 1964) is never used to explainthe persistence of overqualification, except bySichermann (1991), who makes a study of thisaspect of the overqualification problem (p.114).Since the human capital theory may be de-duced from neoclassical premises (see subsec-tion 3.1 above), Rumberger ’s comment,‘overeducation does not really exist in this con-ception’ (Rumberger, 1981b, p.24) applies to thetheory of human capital too.

Nevertheless, the great versatility of the hu-man-capital theory surfaces once again inpartial analyses in the field of overqualifi-cation research when, for example, it is dem-onstrated in western Germany that, in thecourse of a career in which a person’s jobsgenerally match his or her qualifications,transitional periods spent in overqualificationresult in less erosion of human capital – andhence of subsequent earning power – thanequivalent periods of unemployment (see sub-section 6.6 below).

The assignment theory

The assignment theory, which was establishedby Tinbergen (1956), following preparatorywork by Roy (1951), and was further devel-oped by Sattinger (1975), Rosen (1978),Hartog (1980, 1981a and 1981b) and Mac-

14 It should be noted, however, that the employer’sassessment at this stage is based purely on paperqualifications. This systematically ignores the roleof ‘soft skills’, to which increasing importance isattached (see for example Blaschke, 1987). Thereis much evidence to suggest that such qualities –especially in the service sector – are often a neces-sary, though not sufficient, condition for a success-ful career (in the case presented here, professionalexpertise – normally certified by a formal qualifi-cation – would also be essential). It is not incon-ceivable, however, that professional expertise and‘soft skills’ could be alternatives. This would meanthat a job applicant’s soft skills could compensate,at least in part, for a lack of professional expertiseand/or paper qualifications.

Felix Büchel

464

Donald (1982) proceeds from the assumptionthat people with different skill levels are as-signed to jobs with different requirement lev-els. Given the complexity of this allocationprocess, mistakes are inevitable. In otherwords, the phenomenon of underskilled (andoverskilled) employment is inherent in anintricately structured market economy. Theincome effect of such mismatches is examinedin assignment literature, which is able to dem-onstrate that the income differential arisingfrom the discrepancy between individuals’formal qualifications and their job descrip-tions is exclusively incurred by neither theemployer, as the job-competition model mightlead us to expect, or by the employee, as thehuman-capital theory would suggest, but isactually divided between both parties (Har-tog, 1986). Hartog shows this explicitly for acase of directly measured overqualification(Hartog, 1985a). The ratio in which this in-come effect is split between the two partiescan also be established for Germany and othercountries.15 When the income losses incurredby each party from a mismatch are reduced,this eases the pressure on either side to avoidoverqualification and makes the phenomenonall the more persistent.

Filtering, signalling and screeningapproaches

Filtering, signalling and screening ap-proaches16 all proceed in similar ways fromthe assumption that human capital createdin formal education does not directly deter-mine employees’ pay levels. The exact futureproductivity of job applicants is not recognis-able to a prospective employer, who faces theproblem of ‘hiring as investment under un-certainty’ (Spence, 1973, p.356). The qualityof the applicants’ training certificates servesas a manifest indicator of the latent variable‘future productivity’.

This resolves the crucial problem of uncer-tainty or imperfect information, since candi-dates with a higher or more relevant educa-tional qualification are expected to be moreproductive than others. The certificate onlyacts as a signal, a screening mechanism anda filter in the recruitment situation. Once thenew recruit has started work, he or she canbe extensively tested, as can the efficiency ofthe basis on which the recruitment decisionwas made, in other words the approaches weare discussing here; the effect of the certifi-cate diminishes rapidly: ‘To the extent thatthe employer does filter and does so accu-rately, the value of the college filter is reduced’(Arrow, 1973, p.215). Even if some certificatesand/or their holders prove unproductive whenput to the test, this will scarcely change em-ployers’ decision-making criteria, since em-ployees with higher educational qualificationsare more productive on average than theirless-educated colleagues: what counts is thelevel of expectation. So it is not the actualhuman capital that yields an investment divi-dend but the certificate. To put it another way,the certificate, being a manifest quantity, is atangible indicator of the holder’s human capi-tal, which, being a latent quantity, is not di-rectly measurable – at least not in the recruit-ment situation; presumably these twoquantities are not perfectly equal. On thebasis of these premises too, the return on in-dividual certificates may be expected to fallwhen the market experiences a glut of labour;nevertheless, the fact that a certificate attest-ing to a higher level of education is not onlydirectly rewarded but also indirectly, becauseit serves as a key to skilled employment, tendsto increase demand for higher qualificationswithin the education system, irrespective ofmarket demand for labour.

This growing demand is further stimulatedby factors extraneous to the labour market,namely the non-monetary returns to highereducation, such as the accretion of social sta-tus and prestige; here too, the educationalcertificate serves as a signalling device.17 Ifthe demand for labour with the skills and

15 See Rumberger, 1987 (United States), Daly etal., forthcoming (United States and Germany),Alba-Ramírez, 1993 (Spain) and Kiker et al., 1997(Portugal).

16 See Arrow, 1973; Spence, 1973 and 1974;Taubman and Wales, 1974, pp.153 ff.; Stiglitz,1975.

17 On this aspect of formal qualifications, seeSmith, 1986, p.98.


465

qualifications in question cannot keep pacewith this expansion, the result – once all thematched posts have been filled– will be a risein the unemployment rate for people withhigher qualifications and/or an increase inoverqualification.18

The distinctive feature of the aforementionedapproaches is that they are almost impossi-ble to test empirically within a microeconomicframework. Although nearly all the relevantempirical studies on the subject of overquali-fication resort to these approaches when theyadvance their explanatory hypotheses, theyoffer no explicit test. They circumvent themeasurement problem by relying on the factthat the empirical legitimacy of these ap-proaches is generally held to be irrefutable.

This probably applies even more to Europeancountries with sophisticated systems of voca-tional education than, for example, to theUnited States. In Germany, for instance, vir-tually every job offer – at least above a mod-est skill level – stipulates that applicantsshould possess a specific formal qualification.The function of the required certificate as asignal and a screening device is not alteredby the fact that, in recent times, especially atthe highly skilled end of the jobs market, mereevidence of vocational training is no longerenough to secure a job. Additional tests, suchas those conducted in the assessment centresof large companies, seem to be a response tothe appreciable weakening of the signal emit-ted by university degrees, for example. Theimplication – which may not be entirely un-founded – is that we are seeing a decline inuniversities’ ability to measure their students’skills and to translate the results of theirmeasurements into a valid form of certifica-tion when students complete their degreecourses. The signal function of a certificate,however, is unquestionably preserved by vir-tue of the fact that, while it may not be a suf-ficient qualification in itself, in the vast ma-jority of cases it is probably at least a sine

qua non for the jobs for which certificate hold-ers apply.19

Job-competition model

The job-competition theory is based on thepremise that an individual’s place in the jobqueue is largely determined by the nature ofthe underlying variable ‘education’. A higherlevel of education – whether or not the jobactually requires it – implies lower trainingcosts for employers. It will likewise take lesstime and effort to familiarise individuals witha job for which they are overqualified thanfor one that matches their formal qualifica-tions.

Still greater importance attaches to the train-ing costs involved in upgrading employees’skills and qualifications. The preference is toinvest in employees who, on account of theireducational background, are likely to be thecheapest to train. Accordingly, it is theoverqualified candidates who raise the high-est expectations in terms of returns on fur-ther educational investments and who there-fore stand the greatest chance of being senton training courses; these corporate expecta-tions and the strategy of externalising train-ing costs to which they relate (in the case ofoverqualified employees, the education sys-tem has already footed a good part of the bill)put overqualified people in pole position inthe labour queue, ahead of those whose quali-fications match the job description; in short,overqualified people have a head start in therecruitment stakes (Thurow, 1975, pp.75ff.,pp.86 et seq. and pp. 91 ff.).

This model is therefore able to explain anemployer’s motivation for preferring over-qualified candidates and, by the same token,the persistence of overqualification.

An important point is that the candidates’formal qualifications only determine theirposition in the job queue. They do not, how-ever, affect the income they can expect to earn;that is determined by the job description

18 For detailed treatment of this phenomenon, seeRumberger, 1981b, pp. 25ff., as well as Tsang andLevin, 1985, p.95.

19 See also Rippe, 1984, pp.76 ff., on this point and,for a more general analysis, Weißhuhn, 1978.

Felix Büchel

466

alone.20 So all that Thurow’s job-competitionmodel directly explains is the motivation ofemployers to put cost and productivity con-siderations first and opt for the candidate withthe best formal qualifications.

This does not directly explain the incentivefor the candidate to accept an underskilledjob, because he or she cannot expect an edu-cation premium. Indirectly, though, the de-sire to acquire the highest possible level ofeducational certificate in order to secure thebest possible position in the job queue helpsto perpetuate overqualification from the sup-ply side. If overqualification is always an op-tion, this strategy seems rational, irrespec-tive of the structure of demand for labour. Thedemand for education in this case is not basedon absolute expectations but is governed bythe desire to secure the best possible relativeposition in the labour market, even at the riskof subsequent devaluation of qualifications.

The upshot of this is that people with higherqualifications are squeezing the less qualifiedout of the jobs they have traditionally occu-pied.21 This applies at least to jobs with morethan minimal skill requirements (cf. van Oursand Ridder, 1995). In this respect, the job-com-petition model is therefore compatible withthe approaches described on the precedingpages.22

Rumberger writes, ‘[...] the model does offeran intuitively appealing interpretation of thephenomenon of overeducation’ (Rumberger,1981b, p.29); this may be the reason why thisis the most frequently cited approach in theempirical literature on overqualification.

There is, however, one serious problem in re-lation to empirical testing, although it has

surprisingly gone unmentioned in the litera-ture that has been produced to date. This con-cerns the question of the type of training inwhich overqualified employees are expectedto enjoy comparative advantages over theircolleagues with well-matched jobs and lessschooling. Is it initial training, when they arenew to their jobs, in other words the trainingof recruits for their posts? Or do these advan-tages relate to future training courses to up-grade their skills and qualifications, particu-larly in connection with career developmentwithin the company? Thurow is not at all spe-cific on this point.

The classic test of the job-competition modelin relation to overqualification involves a com-parison between the training activities under-taken by employees in underskilled jobs andthose undertaken by employees in appropri-ate jobs (see for example Groot, 1993a andHersch, 1995). If the validity of the initialpremise is to be proved, it must be demon-strable that overqualified employees undergoless training than those whose jobs matchtheir qualification levels, since people withbetter formal qualifications need shorter fa-miliarisation periods and less training. Inorder to substantiate the second premise,however, we should need to arrive at the op-posite result, since the greater learning ca-pacity of the more highly educated employ-ees will reduce training costs and thereforeincrease the return on investments in furthertraining; at constant required skill levels,employees in underskilled jobs are thereforemore likely to be sent for training than theirless-educated colleagues whose jobs and quali-fications are well matched.

There are several reasons why the second in-terpretation appears more credible. In veryundemanding jobs, which would be under-skilled for any employee, familiarisation nor-mally takes the form of learning by doing. Atleast in European surveys, however, most ofthe training reported by overqualified re-spondents is formal in nature. If this line ofargument is pursued, the training activitiesreported in surveys must relate primarily tofurther training designed to upgrade skillsand qualifications. Such an interpretationalso draws an analogy with the theory of ca-

20 The same applies to non-monetary returns (seevon Henninges, 1996, p.81).

21 For a more detailed treatment of this point, seeRumberger, 1981b, pp.27 ff., as well as Tsang andLevin, 1985, p.95).

22 For a more extensive discussion of this aspect,including employers’ productivity considerations,see point 2.3.4 above.


467

reer mobility (see below). This not only con-cerns the employer’s motivation but also thatof the employee: the prospect of being at thefront of the queue for further training andthus of achieving at least limited promotionis likely to increase a jobseeker’s willingnessto accept overqualification.

The job-matching theory

The theory of job-matching, which was devel-oped in the 1970s, may be loosely divided intotwo types of model. In the first category, jobsare understood as ‘pure search goods’. Infor-mation about an alternative job, in which anemployee’s qualifications and the job descrip-tion are likely to be more closely matched, isobtainable at all times. If such an alternativeis available elsewhere, the employee willswitch companies (Burdett, 1978; Jovanovic,1979a; Mortensen, 1988). In the second cat-egory of model, jobs are interpreted as ‘experi-ence goods’: ‘That is, the only way to determinethe quality of a particular match is to form thematch and “experience it”’ (Jovanovic, 1979b,p.973, quoting from Nelson, 1970; for studieson this type of model, see Johnson, 1978;Jovanovic, 1979b; Topel and Ward, 1992).

Length of service with one company is tradi-tionally regarded as a suitable indicator of thequality of a match: ‘Job tenure is my (match)quality indicator under the assumption that“good matches endure”’ (Bowlus, 1995, p.336;for analogous rationalisation, see for exam-ple Schasse, 1991, and Topel and Ward, 1992).The assumption is that a mismatch will beidentified relatively quickly by one of the par-ties, and the employment contract will be ter-minated. A good match, on the other hand,benefits both sides, and the contract willtherefore last longer.

In the context of the present study, it seemslogical to measure the quality of the matchdirectly by reference to the correlation be-tween qualifications and job descriptions.Employment in a job that is commensuratewith one’s level of training is obviously morelikely to prove a good match than overquali-fication. Accordingly, the ‘experience-goods’variant of the job-matching theory is able toexplain the persistence of overqualification on

the basis that, on the one hand, employeesneed a certain amount of time in the initialphase of their careers to find a ‘good’ match.Employers, on the other hand, do not respondquickly enough – if at all – to mismatches,sometimes because of a risk-avoidance strat-egy based on the precept that ‘a bird in thehand is worth two in the bush’, and thereforemiss the opportunity to test whether the em-ployee’s skills are better matched to the re-quirements of a different job and hence thechance of improving their company’s overallefficiency.

In the empirical studies on overqualification,the job-matching theory is only used on theodd occasion as a means of explaining thephenomenon. Sicherman (1991) uses it implic-itly in testing his career mobility model:‘Overeducation might be an indication for abad match in the sense that the worker’s edu-cation might qualify him for a better-payingjob. In such a case, it is likely that eventuallythe worker will change his job, that is, hisoccupation, firm, or both’ (p.105). Sichermantests it explicitly when he compares the jobtenure of employees whose job descriptionsare commensurate with their qualificationsand those who are overqualified for their jobs.His finding that mismatched employees haveshorter job tenures (p.106) is hardly surpris-ing. To the extent, however, that a short staywith a company is a traditional indicator of amismatch, as was mentioned above, this testis tautological in a sense; the best it can do isto verify the validity of the mismatch indica-tor selected by the job-matching theorists,namely job tenure.

Alba-Ramírez (1993) chooses a more effectivetest. He not only uses the duration of jobs,which can be observed longitudinally, but alsohas recourse to the employee’s earlier careerhistory. He examines the connection betweenthe average duration of all previous jobs andthe current job match. The analysis is ex-tended to include two further partial indica-tors: whether the employee has ever changedfirms and whether the employee has morethan five years of seniority in his or her cur-rent job. On this basis, Alba-Ramírez likewisefinds empirical evidence to validate thetheory.

Felix Büchel

468

There are, however, two weak spots in thistrain of thought. The first is that Alba-Ramírez is only able to use prior work expe-rience, a ‘potential’ measure; this is likely tocause serious reliability problems, especiallyin the case of older working women. The sec-ond is that he includes in the calculation theemployee’s seniority in his or her present job,which is a piece of right-censored data at thetime of observation. Where employees haveonly had their present jobs for a short time,moreover, it will not yet be possible to estab-lish a valid correlation between job durationand match quality.23

In the present study, we intend to test theempirical evidence for the ‘experience-goods’version of the job-matching approach evenmore explicitly.

The expectation raised by the theory is clearlythat the match between the demands of thejob and the worker’s skills and qualificationswill be all the more perfect the older and moreexperienced an employee is (Franz, 1991,p.203).

Although the argument that ‘older and moreexperienced’ employees stand a better chanceof being in a well-matched job by virtue of theprogressive improvement in their ability toassess their own career prospects is certainlyplausible, it does raise the question of thevalidity of the personal attributes cited byFranz. Merely growing older is obviously notenough in itself. The second attribute, increas-ing experience, certainly seems to be moreserviceable. If experience is not intended tobe synonymous with age, it is plainly not ex-perience of life but occupational experiencethat is meant here. The theory does not sug-gest that occupational experience, which inthis context means the knowledge accumu-lated by the employee about his or her pro-ductive capacity and the uses to which it canbe put in the labour market, is likely to in-crease through continuous employment in oneand the same job but only through a process

of trial and error in various jobs with severaldifferent employers.

For this reason, it seems advisable to regardthe frequency of job moves as a direct meas-ure of occupational experience in empiricaltests.24 This would be used to check whetheran increase in the number of career movesactually reduces the probability of a subse-quent mismatch between formal qualificationand job requirements. A specific test (Büchel,1998b, p.139) demonstrates that – contraryto theoretical expectations – overqualifiedemployees have a record of considerablygreater career mobility than those in jobs thatmatch their level of training. This finding doesmore to prove the plausibility of the segmen-tation approach (see below), which predictsweak company attachment among employeesin undemanding jobs.

Career mobility theory

The theory of career mobility, developed bySicherman and Galor (1990) on the basis ofRosen’s groundwork (1972), postulates thatthe return on investments in higher levels ofeducation not only manifests itself in higherstarting pay but also in better prospects ofpromotion within a company or of an upwardmove to another company. In other words, itfocuses on income development over a periodof time and may be regarded as a refinementof the human-capital theory. If tests controlfor required skill levels, promotion prospectswill increase with rising education levels.

One proposition (‘Corollary 2’) is as follows:‘Individuals may choose an entry level inwhich the direct returns to schooling are lowerthan those in other feasible entry levels if theeffect of schooling on the probability of pro-

23 Alba-Ramírez, like Sicherman (1991), estab-lishes a strong positive correlation between jobtenure and match quality (pp.272-273).

24 Even within the work of individual authors, thisconfiguration is presented in different ways andhence inconsistently. Hersch (1991), for example,rightly speaks of employees who ‘move into bettermatching jobs’ (p.144), which they clearly cannotdo unless they make a change (such as obtainingpromotion within a company or moving to anothercompany), whereas Hersch (1995) refers generallyto ‘previous work experience’ (p.620), which ismeasured in total years of employment.


469

motion is higher in this entry level’ (Sicher-man and Galor, 1990, p.177).

Sicherman (1991) operationalises this sort ofpromotion as both a ‘move to higher-level oc-cupation’ and the attainment of a ‘higher wagelevel’ (p.109). A motive for accepting anunderskilled job can be deduced from theabove hypothesis: ‘[…] it will be rational forsome individuals to spend a portion of theirworking careers in occupations that requirea lower level of schooling than they have ac-quired. This observation can serve as a par-tial explanation for the phenomenon of‘overeducation’’ (Sicherman and Galor, 1990,pp.177-178; cf. Sicherman, 1987). The higherthe probability of promotion, however, thegreater is the possibility that the employeewill quit if promotion is not approved; this isthe other proposition (‘Corollary 1’) of the ca-reer mobility theory (see Sicherman andGalor, 1990, p.176).

The appeal of the career mobility theory doesnot derive solely from its high degree of plau-sibility but also from the fact that – unusu-ally – it can explain both parties’ motivationfor accepting an individual job/education mis-match by means of the same approach.

Employees forego part of what they could beearning now in exchange for a favoured posi-tion in the promotion queue, thereby invest-ing in increased future earning power. If, af-ter some time in the job, they feel that theiremployer has reneged on the deal, their onlyoption is to move to another company.

For employers this arrangement is enticingin the sense that it gives them an opportu-nity to ‘test’ new recruits, whose high level ofqualification predestines them for managerialfunctions, at a reduced rate of pay over a cer-tain period of time. There is something of ananalogy between this and the training proc-ess under the German Duales System of al-ternating classroom and workplace training,in which the company with which a traineehas been working decides at the end of thetraining course whether to offer the traineean employment contract, in other words to‘promote’ him or her to the status of a skilledworker. A comparable configuration is becom-

ing increasingly commonplace in the academicworld, whereby individuals are not appointedto traditional academic posts until they havecompleted a traineeship in the institution inquestion.25 The ‘course fee’ that the trainee isimplicitly expected to pay for his or her on-the-job training thus assumes the nature ofan investment.

In a subsequent study, Sicherman (1991) wentback to the test of the theory developed inSicherman and Galor (1990) and tailored itexplicitly to the phenomenon of overquali-fication. He found clear evidence, which ishardly surprising, of the validity of the ca-reer mobility theory he helped to construct.Employees in underskilled jobs, he estab-lished, are younger than those in well-matched jobs, have a higher rate of internalpromotion and switch companies more often.

Alba-Ramírez (1993) arrives at similar re-sults. The problem with the approach adoptedin both studies, however, is that, in compar-ing employees in underskilled and well-matched jobs, they operate with a constantlevel of formal qualification but different lev-els of skill requirement. If the structures theyidentify correlate highly with job levels, thefindings will inevitably be misinterpreted,because a causal link will be imputed betweenthe various identified effects and the job/edu-cation mismatch, although it is possible thatthe differences between the two groups areentirely due to the fact that the members ofone group have more demanding jobs thanthose of the other group.

In a test the author conducted on this theory(Büchel, 1998b, pp.140 ff.), allowance wasmade for this possibility by ensuring that,once the conventional qualification-based testhad been completed, the effect of overquali-fication was only investigated for certain jobswhich were as homogeneous as possible interms of required skill levels. The test showsfirst of all that there is no significant differ-

25 ‘Evidently, in many companies, an employee’sprecise job is not determined until he or she hassuccessfully completed the traineeship programme’(iwd, 1994b, p.5).

Felix Büchel

470

ence between the promotion prospects ofoverqualified employees and those of ad-equately educated employees. Secondly, onthe basis of the same investigative strategy,the test actually did indicate a negative cor-relation between the internal promotion of anoverqualified employee and the probability ofthat employee leaving the company, just asthe theory predicts, but it also demonstratedthe same phenomenon for adequately edu-cated employees. These findings clearly do notdeliver a resounding endorsement of the ca-reer mobility theory.

In a more extensive test by Büchel andMertens (2000, and forthcoming), the Sicher-man (1991) findings are replicated with Ger-man data (full-time male employees in west-ern Germany) – with the required skill levelagain being held constant. This test actuallydemonstrates a higher probability of promo-tion into an occupational group with a greatermass of human capital for overqualified em-ployees than for those with adequate educa-tion. This also applies if we pursue the ap-proach adopted by Robst (1995a), whoanalyses changes in occupational statusrather than transfers from one occupationalgroup to another. It has been demonstrated,however, that these findings largely derivefrom blurred distinctions between categoriesof occupational status. If changes in incomeare examined instead, and if the test controlsfor the base effect (income in the base year),it emerges that overqualified employees ex-perience less wage growth than their ad-equately educated colleagues. This finding isconsistent with that of the sociological stud-ies which have established a link betweenbetter job/education matches and brightercareer prospects (see for example Sørensen,1977, and Spilerman and Lunde, 1991, p.716).This is another finding which undermines theattempt made in Sicherman (1991) to applythe theory of career mobility to the phenom-enon of overqualification.

The theory of differential overqualification

A remarkable theoretical approach to the ex-planation of a greater risk of mismatch be-tween formal qualification and job require-ments for married women in restricted

markets was developed by Frank (1978b; seealso Frank, 1978a). He criticised the fact thatthe difference in the incidence of mismatchesbetween male and female employees had al-ways been ascribed to the personal charac-teristics of men and women (see for exampleGwartney, 1970, Cohen, 1971, and Fuchs,1971) and that the residual variance had beenrather feebly explained away as ‘discrimina-tion’ (see for example Oaxaca, 1973).

A theory of differential overqualification maybe outlined as follows. The starting point isan income-maximising jobseeker. The ex-pected rate of pay is governed by the skill levelof the job alone, like the expectations in thejob-competition model and the productiontheory (Frank, 1978b, p.362).

If ‘qualification’ is defined very broadly,jobseekers are, to a certain extent, overquali-fied for every job for which they are eligibleto apply, since they will always possess cer-tain knowledge that is never used in the per-formance of their duties. This overquali-fication gives rise to a wage deduction fromthe ‘optimum’ (purely education-dependent)income which would be obtained if a perfectmatch were to be made (p.362). The economi-cally rational strategy for a single individualis therefore simple: the individual identifiesthat vacancy in the market system for whichhe or she is least overqualified. The samplingdistribution of the degree of qualification de-pends on the total number of vacancies in allmarkets combined;26 this follows from the factthat the mobility of an individual within hisor her own region is not restricted.27 The ex-pected degree of overqualification (and henceloss of pay) for the single searcher approacheszero as the total number of vacancies in allmarkets approaches infinity.

In the case of married couples, the searchproblem is considerably more complex. As-suming an inclination towards paid employ-

26 On the concept of local labour markets, see forexample Topel, 1986.

27 Frank chooses not to include transaction costsin the calculation underlying a migration decision.


471

ment on the part of both spouses, their aimwould be to maximise the joint income. Be-cause of the formal basis of the spouses’ re-spective rates of pay, it is not possible simplyto find the lowest possible degree of combinedoverqualification; consideration must also begiven to the spouses’ education levels, whichmay differ (pp.363-364). The optimisationprocess is further complicated by the assumedcondition that both spouses will have to findwork in the same local job market. Four gen-eral properties of the expectations with re-gard to a married couple are identified(pp.364-365):

i) The expected amount of the added over-qualification degrees of a searching couplewill exceed the expected degree of over-qualification for a single searcher.

ii) The expected degree of overqualification ofa husband will be lower than that of hiswife, if his qualification is higher than her´s(and vice versa).

iii)The expected degrees of overqualificationfor both partners will approach zero as thetotal number of vacancies in the marketsystem approaches infinity.

iv)The expected degrees of overqualificationfor couples are not, as in the single-workercase, independent of the distribution ofvacancies across local labour markets. Theexpected degrees of overqualification in-crease as vacancies are more evenly dis-tributed across local labour markets –always assuming that migration is a possi-bility for the couple.

In this dilemma, the sheer complexity of whichwill overwhelm any couples trying to maxim-ise their household income, Frank (1978b) saysthat the husband takes the initiative and actsin accordance with a ‘male chauvinist familylocation decision rule’ (p.364). He begins byseeking the best possible job for himself oncehe has performed the simple task of identify-ing the vacancy for which he is least over-qualified (see above); in so doing, he also de-termines the local labour market in which bothspouses will work. Once this decision has beentaken, the wife conducts her own individual

search to find the best possible job for herselfwithin that market.

Assuming that the couple are entirely free tomigrate, the crucial point is that the husbandis clearly able to conduct his search across thewhole range of local markets, in other wordsto apply for any suitable job vacancy in anymarket. The wife’s subsequent search, how-ever, is restricted to the local job market se-lected by the man. Since the number of jobvacancies in the local market is smaller – verymuch smaller – than in the global market,the wife may logically be expected to find apoorer match. This disadvantage will surelybe all the greater if the husband finds hisoptimum job in a small local market.

The disadvantaged position of married womenmay be reflected in either of two ways. If thewoman is unable to take up a more attractivejob offer at another location because her hus-band has already found his optimum job lo-cally and is not prepared to move, she is a‘tied stayer’ (Mincer, 1978). If, on the otherhand, the man has decided to accept a lucra-tive job offer far away from the couple’spresent home, his wife is compelled to movewith him, irrespective of the quality of hercurrent job or her employment prospects atthe new place of residence; she is a ‘tiedmover’. So whereas marriage tends to en-hance a man’s status in the labour market,28

the opposite may be expected to apply towomen.

A high degree of empirical plausibility makesthis another persuasive theoretical approach.It does not even rely on the assumption thatthe man will adopt an irrational ‘macho’stance to establish the family location againstthe interests of his partner. Since, in terms ofthe whole population, the average level of for-mal qualification for husbands is always ob-servably higher than that of wives, and sincemarried women work fewer hours on averagethan married men, the use of the ‘male chau-vinist family location decision rule’ shouldgenerally be rational too, given the aim of

28 For a summary of the various reasons for this,see Cornwell and Rupert, 1997.

Felix Büchel

472

maximising the total household income – es-pecially as it seems impossible to fully recon-cile both sides’ interests, not only because ofthe mathematical complexity of the decision-making operation but also because of crucialinformation gaps.

When Frank tested the theory (Frank, 1978b),the test was not explicitly tailored to over-qualification but took the form of an incomeanalysis.29 Studies have traditionally focusedon the ‘tied stayer’ variant, presumably be-cause of data limitations.

An income-based analysis, however, raisessignificant methodological problems. If anunequivocal causal link is to be establishedbetween the phenomenon of differentialoverqualification and income differentials, itwill be necessary to control for those income-level variations between larger and smallerlocal labour markets (the usual distinction isbetween municipal and regional markets)which are not productivity-dependent. Suchvariations include, for example, local weight-ing allowances designed to compensate forhigher housing costs. In addition, there wouldbe a need to control for differences betweenmunicipal and regional occupational struc-tures, for instance by including several hun-dred occupational dummies in the income es-timates. None of the studies that haveappeared to date have come up with a satis-factory solution to this problem.

A less complex approach can be adopted if theexamination of relative incomes (which areonly used, after all, as indicators of the job/education mismatch that this theory studies)is replaced by direct examination of the mis-match, which might be done by measuring thedegree of under- or overqualification. The onlyexample of this line of enquiry is found inMcGoldrick and Robst (1996),30 who concludedthat there was no empirical evidence to sup-

port the theory of differential overquali-fication, since the dummy variables they in-corporated to denote different sizes of regiondid not have a significant effect in the modelused to establish whether the probability of amarried women being overqualified for her jobwas greater in smaller than in larger mar-kets (p.282).

This test was reproduced in Büchel, 1998b,p.143 ff.31 The author’s study was also the firstto include tied movers in such a test. It alsotook heed of the criticism made by McGoldrickand Robst (1996) that the number of job va-cancies was an inadequate indicator of theshortage of jobs since it did not take accountof the number of people seeking those jobs; amore suitable measure of the ‘tightness’ of thelabour market, they said, would be the re-gional unemployment rate. At the same time,for the sake of greater validity, this test in-cluded not only married couples but cohabit-ing couples too; married people not living to-gether, on the other hand, were treated assingle.

Like its predecessors, this study had to adoptan implicitly assumed correlative trend be-tween the number of job vacancies, the totalnumber of jobs and the population of the placeof residence when it tested the theory, al-though it also controlled for the regional un-employment rate (see above). The problem ofequating the place of residence with the joblocation can, however, be overcome by at leasta proxy construction which controls for thedistance from home to work.

The study demonstrated that married womenin rural areas actually do register a signifi-cantly above-average overqualification rate if– and only if – an estimated distance fromhome to work is taken into consideration. Thishigh-risk group can therefore avoid theoverqualification syndrome if they commuteinto larger urban areas. However, where mar-ried women living in sparsely populated ar-eas are unable or unwilling to commute, theprobability that they will not find a well-

29 For similar approaches, see Topel, 1986, andOfek and Merrill, 1997.

30 Teichler (1994), who focuses on Germany, iden-tifies academics who have ‘voluntarily’ chosen un-deremployment in preference to a more appropri-ate post because they wanted to be close to theirpartner (p.30).

31 See also Büchel, 1999d, and Büchel (forthcom-ing).


473

matched job in the vicinity of their home isatypically high. This finding accords with theexpectations of the theory.

The job-search theory

The basic assumption of the job-search theory(Stigler, 1961)32 is that jobseekers adopt asearch strategy which will maximise their po-tential lifetime earnings. In the basic model ofthe theory, which has subsequently been ex-tended in many different ways, it is assumedthat, from a density function of wage offers,postulated as a known quantity, the jobseekerchooses the number of offers that will optimisethe job search process. Once this chosennumber of job offers has been reached, thejobseeker will select the highest wage on offer.The longer the jobseeker waits, the greater hisor her chance of receiving an even higher wageoffer. However, the longer the duration of thejob search, which equates to a period of unem-ployment, the higher the cost to the jobseeker,not only in terms of higher direct expenditurearising from job applications but also in termsof opportunity costs, in other words the incomethe jobseeker has foregone in order to wait fora better offer, minus any unemployment ben-efit he or she might have received in the in-terim period. There is also a danger that ac-quired skills will become increasingly obsoleteduring that period.

As the job search goes on, the jobseeker’s tar-get wage may be expected to decrease, sincethe time limit on a person’s working life willcompel the jobseeker to operated within a re-stricted time frame. The longer the search forwork lasts, the less will remain of thejobseeker’s working life. After the wage level,this is the second decisive parameter govern-ing the maximisation of lifetime earnings. Thejob search is maximised if the marginal costof seeking a job is matched by the marginalincrease in future earnings.

The expected decrease in the target wage asthe off-the-job search continues may meanthat the job-search theory can help to explain

why people accept underskilled jobs, becausea lower wage is likely to equate with a lowerrequired skill level, for which the jobseekerwill often be overqualified. The findings con-tained in Büchel (1992), for example, showthat a high percentage of skilled workers inWest Germany can only escape from a lengthyspell of unemployment by accepting jobs inwhich their specialised skills are not required.

It may therefore come as a surprise to dis-cover that, in the existing body of literatureon overqualification, the job-search theory ishardly ever cited as a means of explaining thephenomenon. One implicit exception isPatrinos (1995), who explains the acceptanceof overqualification as a response to abnor-mally heavy pressure to find work but gener-ally focuses on the economic resources at thedisposal of jobseekers.

With regard to the serviceability of the job-search theory in the context of overquali-fication, however, it should be noted that thetheory can only explain the temporary, short-term occurrence of overqualification, since itexplicitly relates to off-the-job searches only.The theory is therefore unable to explain whyoverqualified people, having taken unde-manding jobs, are then unsuccessful in theiron-the-job search for a better job, which inthis context would mean a job that is com-mensurate with their qualifications (on thisexpectation, see Rosenfeld, 1992). Hence,since the theory can only explain mismatchescaused by frictional loss, it is not capable ofexplaining the persistence of a high percent-age of overqualified employees in the labourforce.

3.3.2 Unorthodox approaches

Rumberger (1981, pp.29ff.) speaks of two un-orthodox approaches to the explanation ofpersistent overqualification. One of these isthe segmentation approach, and the other isa ‘radical’ approach; in American terminologyin this field, ‘radical’ stands for ‘Marxist’.

Segmentation approach

The segmentation approach rejects the homo-geneous neoclassical model of the labour mar-

32 For a review of the most important refinementsof this theory, see Lippman and McCall, 1976 a, b.

Felix Büchel

474

ket as unrealistic and postulates the exist-ence of labour-market segments in whichwidely disparate working conditions obtain.The assignment of labour to these marketsegments is decided by employers on the ba-sis of a ‘discriminatory’ weighting of indi-vidual screening indicators, such as formalqualification; it is asserted that opportunitiesto move from one segment to another are ex-tremely limited.

In the original model presented by Doeringerand Piore (1971), the basic assumption wasthat the U.S. labour market is essentially adual market. The primary market is chieflyreserved for the traditional core workforces,and jobs in that market are characterised byhigh wages, employment stability and goodcareer prospects. In a secondary or periph-eral labour market with unskilled, poorly re-munerated and insecure jobs, employers re-cruit their fringe workforces in times ofprosperity; when cyclical downturns comealong these fringe workers can be fairly eas-ily offloaded with minimal transaction costs,because they are loosely attached to the com-pany, which has accordingly invested little inthe development of their human capital33.

This simple methodological concept hasproved to be highly resilient. Individual re-finements, such as attempts to take accountof regional characteristics, have seldomgained general currency. In the West Germanlabour market, for example, Lutz andSengenberger (1974) claimed to have identi-fied a separate segment of the ‘specialisedlabour market’ which was characterised byinsecure conditions of employment for skilledworkers with good formal qualifications. Thisregional distinction was intended to take ac-count of the fact that the Federal Republic ofGermany, unlike the United States, had anestablished system of alternating on-the-joband theoretical training.

This distinction, however, is open to criticismon the grounds that every job requires a cer-tain degree of skill, however minimal. Whysuccessful completion of an apprenticeship,

of all things, should be used as the decisiveskill-categorisation criterion largely defieslogic. The skill level required for a job is tra-ditionally operationalised on the basis of thewage paid for that job, a variable that ishardly likely to be governed by ‘natural’thresholds such as the difference between thewage of an unskilled or semi-skilled workerand that of a skilled worker. The infinite ar-ray of criteria that could be selected to distin-guish between segments of the labour mar-ket and the resulting diversity of typologicalsegmentation, which could hardly stand upto a validity test, have already come in forcriticism from several sources, and rightly so(cf. Blien, 1986, p.147).

The segmentation approach does help to ex-plain the persistence of overqualification inconjunction with the screening approach andthe job-competition model. If all suitable jobsin the primary segment of the labour marketare initially occupied, jobseekers who are lesswell equipped to withstand the screening de-vices (applicants with lower pass gradeswhere formal qualification is the screeningdevice, for example) have to settle for jobs inthe peripheral segment. Since the boundarybetween the segments tends to be impervi-ous, the result is a state-dependence effect,whereby the fact that a person works in theperipheral segment becomes a further nega-tive screening device (see also Rumberger,1981b, p.32).

Neubäumer, 1993 (see also Neubäumer, 1997,and forthcoming), adds a noteworthy dimen-sion to conventional consideration of marketsegmentation in the present context by pos-tulating the segmentation not only of the la-bour market but also the market in courseplaces within the system of vocational train-ing. In those occupations and industrieswhere training and working conditions arepoor (e.g. for hairdressers), training is pro-vided in excess of requirements – not onlybecause the cost-benefit ratio of training isespecially favourable to employers in theseareas of activity but also – and this is the nubof Neubäumer’s argument – as a means ofcoping with the loss of the best trainees, whomove upmarket when they have served theirapprenticeship. Because even the training33 For an overview, see Cain, 1976.


475

market is segmented, the less gifted appli-cants for training places, even though theyare aware of the aforementioned situation, arecompelled to accept training places in the pe-ripheral segment of the training market. Inso doing, they are also running the risk, oncethey have completed their course, of neitherfinding an appropriate job in the occupationfor which they have trained nor of moving intoa better segment of the market (assumingthey are prepared to switch to another occu-pation). In such circumstances, their only al-ternative is often to settle for a relativelymenial job.

Because its underlying hypothesis possessesa certain plausibility, the popularity of thesegmentation approach remains undimin-ished. In very specific areas of the labourmarket, it is certainly possible to observestructures that are consistent with the propo-sitions of the segmentation approach (see forexample Büchel, 1994a). Economics-basedanalyses of overqualification, however, rarelyhave recourse to the segmentation approach.One reason may be that the approach cannotbe built on theoretically self-contained foun-dations. The propositions of the segmentationapproach can be easily constructed in a par-tially analytical manner on the basis of hy-potheses deriving from various microeconomictheories, such as the screening approach orthe job-competition model34 – but this alsomeans that supposedly ‘segmented’ structuresin the labour market can be explained in eco-nomic terms without recourse to the segmen-tation approach.

The Marxist dialectical approach

From a Marxist perspective, the incidence ofoverqualification on a large scale serves theinterests of capital. The phenomenon there-fore performs a similar structural function tomass unemployment, in that overqualifiedemployees are regarded as a reserve armywhose members could easily be reactivated(i.e. drafted into more demanding jobs) if theneed arose. As this reserve of skills grows in

size, the labour force comes under increasingpressure to ‘toe the line’, while wages are sub-ject to downward pressure (for more precisedetails, see Rumberger, 1981b, pp.32 ff.; for abrief outline, see Levin, 1995, pp. 14-15). Thesystematic overproduction of qualifications inthe education system (which can be manipu-lated by the capitalists in a capitalist statemonopoly) therefore serves the purpose ofsystematically weakening the working classin the never-ending class conflict (cf. Bowlesand Gintis, 1976, quoted in Rumberger,1981b, p.35).

This view, however, is invalidated by the factthat the systematic expansion of educationprovision and the associated desire for an endto elitist educational privilege for the upperclasses were being consistently advocated inthe sixties by supporters of the far Left too.

Even as recently as the mid-eighties, voicesin West Germany were calling for the prob-lem of the overqualification of highly quali-fied labour and the question of the wherea-bouts and professional situation of universityand college graduates to be ‘addressed interms of class theory’ (Krais, 1983, p.36).

Although there are unmistakable signs insome European countries that employers mayindeed be deriving certain benefits from thecurrent mass unemployment, such as the re-sultant erosion of trade-union power, thisalone is scarcely able to explain satisfactorilywhy the phenomena of unemployment andoverqualification are so persistent. A disin-clination on the part of employers to take ef-fective action to eliminate such inefficientdeployment of human capital, even if such anassessment could be substantiated, would beeasily explicable within the framework of anempirical analytical theory, because it is nomore than economically rational behaviour.A connection between overqualification andclass conflict seems to be even more spurious.

The Marxist dialectic approach to economicshas, of course, lost much of its appeal overthe last few years. The sole purpose of describ-ing it here is to ensure that all the relevanttheories are chronicled in this literature sur-vey for the sake of completeness.34 See for example Hartog, 1985a, p.281.

Felix Büchel

476

3.4 Extraneous conditions

3.4.1 Institutional regulation

The methods based on partial analysis whichare used to explain the persistence ofoverqualification and which were outlined inpoint 3.3.1 above are abstracted from the con-ditions imposed by the institutional frame-work. In addition to the indirect effect of theseconditions, it is likely that their practical ap-plication will have a direct impact too. In thecountries of Europe with their highly devel-oped welfare systems and their traditionallyhigh degree of corporatism, such conditionswill probably have a far greater effect than,for example, in the United States, where mostof the aforementioned approaches were devel-oped. Remarkably, scarcely any of the relevantempirical works examine this aspect ofoverqualification.35

One important factor in people’s willingnessto accept overqualification is likely to be thedifferential between the rate of unemploy-ment benefit or, alternatively, social securityand the rate of pay for underskilled (normallysimple) work. This brings two regulatory ar-rangements into play – minimum wages se-cured by the trade unions and the rate of un-employment benefit and social securitypayments set by the government. In recenttimes, as the ‘jobs miracle’ has unfolded inthe United States, more and more voices in anumber of European countries have been criti-cising the excessively narrow gap between theminimum wage and welfare benefits. As thegap narrows, the volume of overqualificationmay be expected to fall, because – in the eyesof many jobseekers – the reward for economicactivity is not the pay itself but only theamount by which the wage exceeds the rateof welfare benefit. It does not ‘pay’, in otherwords, to take a cheap job.

The employment authorities are trying tocounteract these calculations by enforcing thestatutory criteria by which a reasonable job

offer is defined. As these criteria become in-creasingly rigid, sometimes even compellingan unemployed individual to chose betweenunderskilled work and loss of benefits, we mayexpect the percentage of overqualified employ-ees to rise.

In Germany, for instance, these criteria haverecently been considerably tightened in re-sponse to the sustained rise in unemploy-ment. Staff directive 30/90 of the FederalEmployment Services (Bundesanstalt fürArbeit), dated 19 March 1990, rescinded oneof the provisions of staff directive 100/82 of13April1982, which laid down that place-ment in employment at a lower skill levelshould only be effected if, despite adequateand appropriate efforts to fill the post inquestion over a period normally lasting threeweeks, no jobseeker with the required lowerskill level could be recruited.

On 1 April 1997, the new section 103b of thePromotion of Employment Act (Arbeitsförde-rungsgesetz) entered into force, drasticallycurtailing once more the right of unemployedpeople to reject job offers. In the first threemonths of unemployment, offers of jobs pay-ing up to 20% less than the jobseeker’s previ-ous employment are classed as reasonableoffers, and in the following three months thisfigure rises to 30%; from the start of the sev-enth month, a job offer is considered reason-able if the net wage (minus necessary employ-ment-related expenditure) is at least equal tothe rate of unemployment benefit (paragraph3); in the case of full-time work, commutingtimes of up to three hours per day are consid-ered reasonable (paragraph 4), as are tempo-rary separation of spouses and ‘Occupations[…] for which the employee is or is not trainedor which he or she has or has not previouslypursued’ (paragraph 5 (my italics); see Fed-eral Law Gazette (Bundesgesetzblatt) 1997/I,No 20 of 27 March 1997, p.700). Althoughthese rules seem highly restrictive at firstsight, the German welfare system is still rela-tively generous by international standards;in the United States, for example, entitlementto unemployment benefit is subject to thestrict proviso that any available job offer –normally at the minimum wage – must beaccepted.

35 Tsang and Levin (1985, p.101) and Rumberger(1981b, pp.124-125) are exceptions, although theirpropositions are couched in very general terms.


477

It is evident that curtailing the right to rejectjob offers will not of itself influence the per-centage of jobs that are occupied by over-qualified people. The impact of such a meas-ure will very much depend on the way in whichjobcentres apply the relevant legislation. InGermany, for example, there are some groundsfor assuming that, even before the latestamendment of the Employment of PromotionAct, the criteria for the definition of reason-able work,36 which were already rigorously for-mulated, were not applied very consistently.

Some evidence for this allegation may be de-rived from the fact that labour offices rarelyorder the suspension of benefits for rejectionof a job offer under section119(1)(2) to (1)(4)of the Promotion of Employment Act (see forexample the official bulletin (ANBA) of theFederal Employment Services, issue No 7/1997, Übersichten II, 7/8, pp.1042-3). Thisfinding may indicate that very few offers ofunderskilled work have been rejected in prac-tice, which would point to the efficiency of thestrict criteria for the definition of reasonablejob offers. But if this is the case, at a timewhen it is difficult to place jobseekers, the ris-ing level of unemployment should be accom-panied by a parallel rise in the percentage ofjobs occupied by overqualified employees; this,however, is not borne out by the availableempirical data (see Büchel and Weißhuhn,1997a and 1998).

The formulation of statutory provisions gov-erning protection against dismissal may becited as another institutionally imposed con-dition that influences the volume of overquali-fication. In this domain too, there are greatdifferences between many European countriesand the United States, for example. It is con-ceivable, for instance, that an overqualified

employee whose company wishes to offloadhim for productivity reasons has no prospectof finding a job commensurate with his levelof training and that he therefore invokes hisright to protection against dismissal. In theUnited States, he could expect dismissal anda place in the dole queue, but in countries withsophisticated systems of statutory protectionagainst dismissal, such systems tend to in-crease the overqualified percentage of theactive labour force.

Another institutional condition which probablyhas a considerable impact on the percentageof jobs occupied by overqualified employees isthe funding structure of the various componentparts of the education system. Higher educa-tion in Germany, for example, is largely state-funded, which is not the case in other coun-tries. The external effects of this fundingstructure grossly distort the situation in theeducation market and are liable to produce asteady flow of overqualified graduates. Whilethe acquisition of educational qualifications atprices ‘below the market rate’ amounts to ra-tional behaviour in the eyes of the individual,since, given the current imbalance betweensupply and demand, highly qualified job-seekers are still in pole position in the labourmarket, in terms of the national economy itgenerates an excess demand for education.

Besides the conditions enumerated above,there are a host of other governmental or semi-governmental regulatory mechanisms whichcan directly affect the qualification structureof the labour force and hence, ceteris paribus,the percentage of overqualified employees.These mechanisms include the setting of qual-ity standards in the education system, the for-mulation of rules governing scholarships andstudent grants and so forth.37 A detailed dis-cussion of every conceivable contributing fac-tor would be an unnecessary digression in thepresent context; the sole purpose of this partof the study has been to draw attention to thepotential impact of the basic conditions inwhich the employment structure operates.

36 See section 1, and more especially section 2(3) ofthe administrative order issued by the FederalEmployment Services (Bundesanstalt für Arbeit)in Nuremberg on 16 March 1982 concerning as-sessment of the reasonableness of a job offer(Zumutbarkeits-Anordnung – documented in thespecial issue of ANBA, dated 15 April 1982). Theauthor wishes to thank Mr Reichel of the FederalEmployment Services for putting together therelevant documentation.

37 See for example Keller, 1997, and Marsden, 1997,on the effects of deregulation.

Felix Büchel

478

3.4.2 The production structure andconditions in the labour market

The changes that the production structureundergoes in the course of time – as a resultof technological progress, for example – canhave a direct effect on the overqualified per-centage of the labour force. For example, theincidence of underskilled jobs may be expectedto rise if qualifications which were once indemand in the labour market (and were there-fore produced by the education system) havebeen devalued over a period of time as theconditions of production have changed. A clas-sic example of this would be the training ofblacksmiths. In addition, job descriptions andskill profiles for particular occupations maychange so dramatically with the passage oftime that even those employees whose quali-fications originally matched their job require-ments will eventually come to realise thatthey are no longer qualified to do their job bypresent-day standards. Such a developmentis observable in journalism, for example.While it is conceivable that further trainingcould update employees’ skill and qualifica-tion levels, horizontal effects, which are notthe subject of the present study, are likely toprove more significant than vertical effects.Irrespective of their empirical significance,however, such structural effects are possibleand should therefore be investigated as ameans of explaining the occurrence of over-qualification.

The findings contained in Büchel (1998b, Ta-ble 5.16, p.216), however, show that this isnot the most common route to overquali-fication. These findings, which were control-led for a number of variables – several socio-economic characteristics of the employeesample and general economic conditions –reveal that overqualification is immediatelypreceded by unemployment, by full-timetraining and by voluntary economic inactiv-ity far more frequently than by adequatelyskilled employment.38

Besides the aforementioned structuralchanges that take place over a lengthy period,more rapid changes in conditions within thelabour market might also conceivably influ-ence the overall balance between the skillprofiles of jobs and those of their incumbents.It may be expected, for example, that contrac-tions and expansions of the labour marketwhich result from cyclical or seasonal fluctua-tions will directly affect the probability ofindividuals being compelled to accept over-qualification, not least on the basis of the hy-pothesis that overqualification is a direct al-ternative to unemployment in many cases.Büchel and Weißhuhn (1998) cite actual evi-dence of dependence – and inverse depend-ence – of the aggregate growth of unemploy-ment on the level of overqualification.

3.4.3 Employees’ utility preferences

The theoretical approaches discussed in point3.3.1 above all assume behaviour patternsbased on income maximisation. If this as-sumption is set aside and utility maximisationtakes the place of income maximisation, newfactors come to light – factors which are rel-evant in the context of overqualification.

In the typical dichotomy between unemploy-ment and overqualification, leisure time39

plays a key role as a preference in a conven-tional assessment of the utility of a job offer.A change in the importance attached to lei-sure time, for example as part of a generalshift in social values or as a regionally (asopposed to nationally, for example) condi-tioned reassessment of leisure time in Europewill therefore directly affect the percentageof jobs occupied by overqualified employees.The linkage is likely to be even less elasticwith unemployment as the alternative thanwith the conventional alternative of a form ofeconomic inactivity other than unemploy-ment, since the ‘distorting’ base effect of un-employment benefit has to be taken into ac-count.

38 The most common access route is from unem-ployment, followed by full-time training, with vol-untary inactivity in third place.

39 It may seem euphemistic to refer to a period ofunemployment as ‘leisure time’ but is neverthe-less based on specialised usage.


479

As for the alternative status of employmentcommensurate with qualification level, whichhas been largely neglected in the discussionof the phenomenon of overqualification (seesubsection 3.2 above), the connections be-tween utility preferences and the decision tobe taken are presumably even more intricate.Individuals who are not out to maximise theirincome have to assess a complex set of jobcharacteristics.40 When people take more de-manding jobs, they can expect heavier work-loads, longer journeys to and from work,greater pressure to work overtime, less scopefor negotiating reduced working hours andmore besides. The ‘purpose’ of a job is also afactor. The need for a sense of purpose issurely universal, but it may take differentforms from one social milieu to another.Schlegelmilch (1987), who reports findingsfrom an analysis of job content among uni-versity graduates who have opted for an al-ternative lifestyle, records the following com-ments from a graduate working as a masseusein a sauna: ‘When I think of all that intellec-tual work, all that studying … it makes mesick. You sit there at your desk, and maybeyou write the odd article, and somebody saysit’s quite good; that happens maybe once ortwice in a year.’ (Schlegelmilch, 1987, pp.179-180). In an earlier work, Schlegelmilch alsolists a number of irrational grounds (from anincome-maximising point of view) for prefer-ring underskilled to adequately skilled em-ployment, irrespective of the demand situa-tion in the market for adequately skilled jobs.These grounds include problems with a pro-fessional role, such as that of a teacher, an-ticipated problems with superiors (encoun-tered in people of an anti-authoritariandisposition), the disillusionment of individu-als with what they could achieve in their oc-cupation in the prevailing conditions and soforth (Schlegelmilch, 1982, pp.418-419).

Individuals who rate the utility of non-mon-etary job characteristics higher than that ofincome are more likely to accept overquali-fication. A change in people’s priorities as part

of a general shift in social values also has adirect effect on the percentage of the total la-bour force in underskilled jobs.

Many of the empirical findings in Büchel(1998b) show that people who subscribe to thestatement ‘A successful career is not so im-portant to me’ are more likely to be inoverqualified for their present jobs. This cor-relation is particularly high in the case of WestGerman women in the middle tier of the quali-fication hierarchy, i.e. holders of vocationaldiplomas (Büchel and Weißhuhn, 1997a, Ta-ble 4-WA, pp.68-69); as was mentioned above,however, this pattern was also demonstratedby Schlegelmilch’s studies of disaffectedgraduates.

3.4.4 Employers’ productivityconsiderations

The most common decision facing employersin the context of overqualification (see sub-section 3.2 above) is whether to select anoverqualified job applicant in preference toone whose formal qualifications are commen-surate with the job requirements. Their deci-sions will be governed by productivity-basedconsiderations. This aspect has already beentaken into account in the discussion of cer-tain approaches in point 3.3.1 above; the in-tention in the following paragraphs is to sub-ject it to more thorough scrutiny.

While the production theory-related effects ofoverqualification in the literature of the Eng-lish-speaking world have already been fairlywidely discussed,41 scarcely any findings onthis subject are available for the Europeancontinent (the few exceptions are referred tobelow).

On this point, the author of the standard text-book for German employment researchers,

41 See for example Rumberger, 1981b, pp.101 ff.,Tsang and Levin, 1985, Tsang, 1987, Tsang et al.,1991, and Hersch, 1991; earlier partial analysesof underemployment are also contained inKornhauser (1965), Vroom (1964), Sheppard andHerrick (1972), Kasl (1974), House (1974), Coburn(1975), Mangione and Quinn (1975) and Quinn andMandilovitch (1975).

40 See for example Lucas, 1977, Sattinger, 1977,and Sattinger, 1980; the latter deals with this is-sue in detail on pp.72ff.

Felix Büchel

480

Wolfgang Franz, writes, ‘If a job has a fixedrate of pay […], the company’s aim will be toidentify the most productive applicant for thatjob’, adding in a footnote, ‘This is not neces-sarily the candidate with the highest (formal)qualification. Companies are rightly hesitantto engage “overqualified” candidates, for ex-ample because any dissatisfaction on the partof such an employee may have an adverse ef-fect on his or her productivity […] and becausethe probability that he or she will not staylong with the company is liable to be high’(Franz, 1991, pp. 211-212). The very wordingof these assertions sounds a distinct note ofcaution, which lends Franz’s statement thecharacter of a supposition rather than a vali-dated scientific finding. This means that itcannot be cited as evidence of such behaviour.

The main dimensions of productivity lossesthat result from overqualification respectivelyovereducation are given in the relevant lit-erature in the English language as a lesserdegree of job satisfaction (Khan and Morrow,1991, Johnson and Roy, 1995, and de Witteand Steijn, 1998), poorer health (Amick andLavis, 1998, pp.26 ff.), a higher incidence ofshirking and absenteeism, narcotics (alcohol)consumption at work and ‘sabotage’ (for ageneral discussion of this point, see Tsang andLevin, 1985; also Haugrund, 1990, p.233).Other factors with relevance to business man-agement are an assumed propensity amongoverqualified employees to change firms morefrequently, which would result in a loss ofhuman capital developed specifically for theneeds of the company as well as creating di-rect transaction costs and restricting partici-pation in further training, thereby impedingthe development of know-how within the com-pany (Beneito et al., forthcoming).

If employers are aware of these risks to pro-ductivity that are inherent in overquali-fication, that knowledge may influence thewage rates they set. Unlike the United States,however, many European countries have spe-cial legislation granting an extensive right ofprotection against dismissal. If shirking byan individual worker is only discovered – as-suming that it is discovered – after the end ofthe initial probationary period, those employ-ers who have overqualified staff on their pay-

roll can only cover themselves financially byconverting part of the basic wage into a pro-ductivity bonus – not just for the individualin question but by way of a ‘collective insur-ance premium’.

Such a strategy, however, depends not onlyon possession of all the relevant informationbut also on the absence of trade-union ‘ob-struction’ of the wage-fixing process. Both ofthese aspects create an incentive for the em-ployer to deal with the equation of gross payand productivity on a one-to-one basis, i.e. inrelation to an individual employee.

An examination of these quite considerableproductivity risks to an employer who recruitsoverqualified personnel (which are confirmedin most of the literature from English-speak-ing countries) raises the question whetheremployers have any interest at all in employ-ing people in jobs for which they are over-qualified. Haugrund (1990) sees one such in-centive (in the technical sector) in theexpectation of above-average performancefrom these employees, along with their ‘natu-ral’ function as a ‘centre of competence’ withintheir respective teams (p.233).

The key to this question, however, is the factthat an employer’s basis of assessment canbe influenced by extraneous factors. A changein the information base, for example, may besuch a factor. It is already patently difficultfor employers to measure their employees’productivity at the best of times (see Boden-höfer, 1984, pp.13ff., and Rippe, 1984,pp.76ff.). The productivity effects of overquali-fication, such as an increased incidence ofshirking, must be even harder to measure. Itis conceivable that administrative and tech-nical progress might serve to enhance thesupervisory mechanisms available to employ-ers (cf. Tsang and Levin, 1985). Whereoverqualification is assumed to be havingadverse effects on productivity which cannotbe nullified by pay cuts, the consequenceshould be a reduction in the volume ofoverqualification.

The general difficulties involved in obtain-ing information (with which Rippe, 1994,deals in detail) might also induce employ-


481

ers, however, to base their decisions on sub-jective experience gathered over a lengthyperiod. If the information situation is ex-tremely bad, they have to rely on ‘prejudices’.It is conceivable that, with a general changein social values (in society’s evaluation of theproductive utility of job-sharing or part-timework, for example), the assessment basis onwhich employers take their decisions willalter.

Such a change may even be caused by a ma-jor shift in the market situation: ‘The higherthe percentage of new recruits who are uni-versity graduates, the more widely an increas-ing percentage of graduates in middle man-agement will be accepted as the norm’(Teichler, quoted in Der Spiegel, 1985, p.47).

In the context of the sea change that the la-bour market is clearly undergoing, this ac-ceptance of higher education as the norm islikely to take root not only among job appli-cants42 but also in the minds of employers. Asimilar effect has been observed in the inverserelationship between the strength of thestigma attached to unemployment and thelength of the dole queue (cf. Omori, 1997). Thistype of shift in values could therefore resultin a sharp rise in overqualification in the fu-ture.43

The opposite effect could be achieved by theexertion of direct political influence on theproduction system. Tsang and Levin (1985,p.101) listed some of the levers that could beused for this purpose, such as tax breaks toencourage the employment of appropriatelyqualified staff, but these levers seem almostunworkable. Indeed, Rumberger (1981b,pp.124-125) had already delivered a pessimis-tic verdict on the power of governments toreduce the degree of overqualification in theprivate sector.

Be that as it may, government measures cer-tainly can be an efficient means of reducingthe level of overqualification in the labourmarket as a whole in the sense that the state,as a major employer, is able to define its ownstaff-selection criteria. The findings set outin Büchel and Weißhuhn, 1997a and 1998,show that the state actually does this verysuccessfully; the percentage of overqualifiedstaff in the civil service is minimal, at leastin western Germany.44 The only snag is thatthis strategy goes beyond the scope of thepresent part of our study, since the public sec-tor is not subject to direct productivity crite-ria.

In the European context, apart from the clas-sic income analyses performed by Duncan andHoffmann (1981), productivity effects haveonly ever been analysed for Germany.Haugrund (1990) devoted a monograph to thesubject. Despite its ambitious and highly de-tailed examination method, scarcely any gen-eral extrapolations can be made from the find-ings of the monograph, because it takes theform of a company study and focuses on a fewtechnical trades. The author has also con-ducted three studies (Büchel, 1999b, whichgoes into greater detail, 1999c and 1999a). Itemerges from these that productivity analy-ses only serve a useful purpose if the job-re-quirement level is kept constant. That is theonly configuration which equates to the choicefacing personnel managers who have to fill avacant post.

The studies conducted in western Germanyshow that in unskilled jobs (in which the over-whelming majority of overqualified employ-ees work) overqualified workers tend to keepbetter health, to change firms less frequentlyand to engage in more further-training activi-ties than their less educated colleagues insimilar jobs; there are no significant differ-ences between the two groups in terms of theproductivity indicators absenteeism and jobsatisfaction.

42 Tsang et al. (1991), for example, demonstratethat the adverse effect of overqualification on jobsatisfaction has been diminished over the years.

43 For a similar hypothesis relating to the academicworld, see Büchel, 1996a.

44 See Büchel and Weißhuhn, 1997a and 1998, Ta-bles 3-W-84, -91, -93 and -95; see also Keller andKlein, 1994.

Felix Büchel

482

This result is consistent with Thurow’s job-competition theory, which predicts that peo-ple with more education will be at the frontof the labour queue. On the other hand, it con-tradicts the traditional expectation thatoverqualified personnel will feel frustratedand therefore prove less productive than theiradequately educated colleagues.

If overqualified workers are less productivethan their adequately qualified counterparts,as is generally reported in the relevant lit-erature, it is solely because these findings areobtained with enquiry methods in which theformal level of qualification is held constantrather than the skill level required for the job.It is obvious that such a strategy can bearlittle fruit, for it is hardly surprising that ataxi driver with a law degree is less produc-tive than a law graduate working as a solici-tor. The only relevant question in this con-text is whether the taxi driver with the legalqualification is less productive than his fel-low cabbies whose only formal vocationalqualification is their taxi driver’s licence.

In general terms, the findings of the author’saforementioned studies can serve to breakdown prejudices about the productivity ofoverqualified employees and go at least someway to explaining the persistence of over-qualification in the labour market by demon-strating that it makes economic sense foremployers to hire overqualified staff, becausethey can be expected to be more productive ina given job than other, less qualified candi-dates.

4. Strategies for measuring theeducational adequacy of a job

4.1 Existing measurement strategies

On the problems involved in measuringoverqualification, the OECD notes in its Em-ployment Outlook that ‘invisible underem-ployment [the term used by the OECD to de-note overqualification] refers to individualswho are working in jobs where their skills arenot adequately utilised, and by its very na-ture is difficult to measure. For this reason,it is not discussed.’ (OECD, 1995a, p.45, au-

thor’s italics).45 What the OECD is express-ing here is that there is no consensus amongemployment researchers on a measurementstrategy, which is why no internationallystandardised tables of comparative nationalstatistics can be compiled for overqualificationas they are for other phenomena, such asunemployment. The following paragraphsprovide information about the various meas-urement issues. The box presents a typologyof different measurement approaches.

Here box4.1.1 The ‘objective’ approach

In the early days of empirical analysis ofoverqualification in the United States, the so-called ‘objective’ DOT/GED approach was re-garded as the standard process for measur-ing the discrepancy between an employee’slevel of educational attainment and the ac-tual education level required for his or herjob. The measurement process is consideredto be ‘objective’, in so far as it does not rely onthe employee’s own subjective assessment ofthe required education level. The basic prin-ciple of the measurement strategy is that itmeasures the required education level for ajob by reference to its occupational category;all important microeconomic details of thisoccupation are recorded. Each occupationlisted in the Dictionary of Occupational Ti-tles (DOT; see Fine, 1968) is allocated a levelof ‘general educational development’ (GED)from a scale of GED values.46 In the secondstage of the process, this educational devel-opment level is translated into an equivalentnumber of years of schooling (see Eckhaus,1964).

45 This means that Rumberger’s postulate in anearlier OECD publication that the model he wasdeveloping was an internationally comparablemeasure of overqualification (Rumberger, 1994,p.281) must have fallen on deaf ears – at least inthe short term.

46 Attempts were made to introduce an alternativescale (‘specific vocational preparation’ – SVP), butit failed to establish a foothold in the field ofovereducation research (cf. Fine, 1968, Scoville,1966, and, for an explicit appraisal, Kalleberg andSørensen, 1973, p.221).


483

qualification among academics in the west-ern part of Germany on the basis of the na-tional microcensus. Since a standardised GEDscale is not available for Germany, the authorsthemselves had assessed each individual jobtype to establish whether it was adequatelyskilled for an incumbent with academic train-ing. This strategy, however, produces an una-voidable ‘grey area’ (Hecker, 1992, p.4) of oc-cupations to which it is impossible to assignan unequivocal education level. Plicht,Schober and Schreyer call this the ‘mixed cat-egory’ of occupations.

A variant combining the approaches describedin the last two paragraphs is described inGroot (1996). Although the occupations of theemployees he studies are listed in the BritishHousehold Panel Survey (BHPS), on the ba-sis of which he conducts his examination, noGED rating is available for the various occu-pations, just as in the case of the Germanmicrocensus. Groot compensates by calculat-ing the mean number of years’ schooling ofthe employees in each occupational category;those whose schooling exceeds the average fortheir group by more than a standard devia-tion are classed as overqualified.

The validity of the DOT/GED methodologyhas been criticised on several occasions, andwith good reason. The first problem lies inthe wide diversity of skill levels required fordifferent jobs within a single occupationalcategory, which the GED system does not takeinto account: ‘[...] estimates of the mean yearsof required schooling in an occupation areconstructed by aggregating jobs, thereby ig-noring variation in the mean years of requiredschooling across jobs within an occupation’(Halaby, 1994, p.48).

Moreover, the use of a one-digit code to evalu-ate the level of education required for a jobclearly cannot reflect the complexity of train-ing-requirement profiles: ‘The GED scores aresimply not detailed enough to produce sensi-tive measures and have validity problems oftheir own’ (Clogg and Shockey, 1984, p.254).Besides, the conversion of GED into requiredyears of schooling is not standardised.Rumberger (1987) rightly expresses the follow-ing criticism: ‘Another problem with the DOT

Once the formal qualification level in the formof the highest reported educational certificatehas also been converted into an equivalent inyears of schooling (unless, as usually happensin the United States, the information has al-ready been collected in the form of a numberof completed years of schooling), the numberof surplus or deficit years of education caneasily be established by subtracting oneamount from the other; thus, overeducation(in years) equals years of schooling completedminus GED levels (in years) (see for exampleRumberger, 1981b, p.58).

The measurement scale for the variables thatare calculated in this way is recognisablybased on the requirements of the human-capi-tal approach. Accordingly, income effects ofsurplus and deficit components of humancapital can be analysed in the framework ofthe traditional evaluation procedures, and thereturns to these components can be comparedwith the returns to the ‘required’ componentsof human capital (see for Duncan andHoffmann, 1981; Kiker et al., 1997, and Dalyet al. (forthcoming)).

Teresa Sullivan (1978), Clogg (1979), Cloggand Sullivan (1983) and Clogg and Shockey(1984) increased the reliability of this meas-urement by excluding from the definition ofoverqualified workers all persons with a sur-plus of zero years of schooling; in their placethey proposed the insertion of a ‘safety mar-gin’ comprising a standard deviation in ex-cess of the mean duration of education usedfor a particular occupation (‘analog for deficityears’); this approach was subsequentlyadopted by other researchers (see for exam-ple Verdugo and Turner Verdugo, 1989, andGroot, 1993a). At this point, one should addthe methodological remark that Hartog (1997,p. 2) and Hartog and Jonker (1998, p. 102)classify this type of measurement as an own,third category apart of the so-called objectiveand subjective approach (see section 4.1.2below). The present author, however, followsthe argumentation of Groot (1996) who takesit as a variant of the DOT/GED approach.

An enquiry strategy derived from the DOT/GED approach can be found in Plicht et al.,1994, who analysed the structure of over-

Felix Büchel

484

Box

: Typ

olog

y of

ove

rqua

lific

atio

n m

easu

rem

ent a

ppro

ache

s

Sel

ecte

d fo

llow

-up

empi

rical

stu

dies

base

d on

app

roac

h

Har

tog

(198

0),

Bur

ris (

1983

(c))

,R

umbe

rger

(19

87),

Pat

rinos

(19

97),

Bat

enbu

rg a

ndde

Witt

e (1

998)

Dis

adva

ntag

es o

fap

proa

ch

Het

erog

enei

ty o

f job

sw

ithin

an

occu

patio

n is

negl

ecte

d

New

occ

upat

ions

, suc

has

e.g

. in

the

IT in

dus-

try,

can

not b

e co

nsid

-er

ed (

until

a n

ew D

OT

isre

alis

ed).

Cha

nge

in jo

b re

quire

-m

ents

is n

egle

cted

(un

tila

new

GE

D s

cale

isre

alis

ed).

Def

inin

g re

quire

d sk

illle

vels

(co

nstr

uctio

n of

GE

D s

cale

) is

met

hodo

-lo

gica

lly p

robl

emat

ic.

Mak

ing

the

GE

D s

cale

com

patib

le to

the

scal

eof

acq

uire

d sc

hool

ing

ism

etho

dolo

gica

llypr

oble

mat

ic.

Usu

ally

hig

h nu

mbe

rs o

fm

issi

ng v

alue

s in

occu

patio

nal i

nfor

ma-

tion

(cau

sed

by c

odin

gpr

oble

ms)

.

Adv

anta

ges

of a

p-pr

oach

No

nega

tive

influ

ence

sin

the

mea

sure

men

tpr

oces

s w

hich

are

caus

ed b

y su

bjec

tive

asse

ssm

ents

of j

ob-

hold

ers.

Dev

elop

ers

of a

p-pr

oach

and

ear

lyst

udie

s

(Pre

limin

ary

wor

k:E

ckau

s 19

64,

Fin

e 19

68)

Sco

ville

(19

66),

Ber

g (1

970)

,K

alle

berg

and

Sø

rens

en (

1973

)

Sho

rt d

escr

iptio

n of

app

roac

h

Ia:

‘Obj

ectiv

e’ a

ppro

ach

For

eac

h oc

cupa

tion

liste

d in

the

Dic

tiona

ry o

foc

cupa

tiona

l titl

es (

DO

T),

info

rmat

ion

abou

t the

leve

lof

edu

catio

nal r

equi

rem

ents

is a

vaila

ble

in a

nor

dere

d sc

ale

(Gen

eral

edu

catio

nal d

evel

opm

ent ,

(GE

D))

.

The

GE

D s

cale

has

to b

e m

ade

com

patib

le to

the

scal

e on

whi

ch in

divi

dual

s re

port

thei

r ac

quire

dsc

hool

ing

(e.g

. tra

nsfo

rmat

ion

into

req

uest

ed y

ears

of s

choo

ling)

.

The

info

rmat

ion

abou

t the

req

uest

ed s

choo

ling

tope

rfor

m a

spe

cific

occ

upat

ion

(SR

) is

com

pare

d w

ithth

e ac

quire

d sc

hool

ing

(SA

) of

job-

hold

ers.

Ove

rqua

lific

atio

n of

an

indi

vidu

al is

sta

ted,

if S

A >

SR

.


485

Box

: con

tinue

d

Sel

ecte

d fo

llow

-up

empi

rical

stu

dies

base

d on

app

roac

h

Sho

ckey

(19

89),

Ver

dugo

and

Tur

ner

Ver

dugo

(19

89),

Gro

ot (

1993

(a),

199

6),

Kik

er e

t al.

(199

7),

Alp

in e

t al.

(199

8),

Coh

n an

d N

g (1

999)

,C

ohn

et a

l. (1

999)

,M

ende

s de

Oliv

eira

et

al. (

fort

hcom

ing)

Dis

adva

ntag

es o

fap

proa

ch

Gen

uine

pro

blem

of

negl

ectin

g he

tero

gene

-ity

of j

obs

with

in a

noc

cupa

tion

rem

ains

.

Met

hodo

logi

cal p

rob-

lem

s in

mea

surin

gm

eans

med

ians

, or

mod

es, a

nd e

spec

ially

mea

surin

g st

anda

rdde

viat

ions

, in

occu

pa-

tions

with

few

em

-pl

oyed

.

Pro

blem

s w

ith c

ausa

lre

latio

ns: t

he h

ighe

r th

esh

are

of ‘e

ffect

ivel

y’ov

erqu

alifi

ed w

orke

rsw

ithin

a s

peci

fic o

ccu-

patio

n, th

e lo

wer

the

mea

sure

d sh

are

‘det

ecte

d’ o

verq

ualif

ied

wor

kers

(an

d vi

ceve

rsa)

.

Adv

anta

ges

of a

p-pr

oach

Hig

her

valid

ity w

hen

cate

goris

ing

over

qual

ifica

tion.

Dev

elop

ers

of a

p-pr

oach

and

ear

lyst

udie

s

Sul

livan

(19

78),

Clo

gg (

1979

),C

logg

and

Sho

ckey

(19

84)

Sho

rt d

escr

iptio

n of

app

roac

h

Ib:

Var

iant

of

‘obj

ectiv

e’ a

ppro

ach:

‘rea

lised

mat

chap

proa

ch’

With

in e

ach

occu

patio

n (d

isag

greg

ated

as

low

as

poss

ible

, usu

ally

on

a 3-

digi

t-le

vel),

the

mea

n,m

edia

n, o

r m

ode

(M)

and

the

stan

dard

dev

iatio

n(S

D)

of a

cqui

red

scho

olin

g of

job-

hold

ers

is c

alcu

-la

ted.

Info

rmat

ion

abou

t M a

nd S

D o

f a s

peci

fic o

ccup

atio

nis

com

pare

d w

ith th

e ac

quire

d sc

hool

ing

(SA

) of

job-

hold

ers.

Ove

rqua

lific

atio

n of

an

indi

vidu

al is

sta

ted,

if S

A >

(M

+ S

D).

Felix Büchel

486

Box

: con

tinue

d

Sel

ecte

d fo

llow

-up

empi

rical

stu

dies

base

d on

app

roac

h

Sic

herm

an (

1991

),Ts

ang

et a

l. (1

991)

[with

an

exte

nsio

n us

ing

App

roac

h Ia

],H

ersc

h (1

991)

,A

lba-

Ram

írez

199

3,H

ersc

h (1

995)

,Jo

hnso

n an

d Jo

hnso

n(1

995)

,R

obst

(19

95(b

)),

McG

oldr

ick

and

Rob

st(1

996)

,S

loan

e et

al.

(199

6),

Bor

ghan

s an

d S

mits

(199

7),

Bat

tu e

t al.

(199

8(b)

),B

orgh

ans

et a

l. (1

998)

,B

attu

et a

l. (1

999)

,va

n de

r V

elde

n an

d va

nS

moo

renb

urg

(199

9),

Ben

eito

et a

l.(f

orth

com

ing)

,D

olto

n an

d V

igno

les

(for

thco

min

g),

Vah

ey (

fort

hcom

ing)

Dis

adva

ntag

es o

fap

proa

ch

Sub

ject

ive

influ

ence

inm

easu

rem

ent a

p-pr

oach

: e.g

. ris

k of

getti

ng a

nsw

ers

influ

-en

ced

by a

cog

nitiv

edi

sson

ance

beh

avio

ur;

answ

ers

may

be

bias

edto

war

ds p

ompo

usne

ssor

exa

gger

ated

mod

-es

ty.

Adv

anta

ges

of a

p-pr

oach

Sev

ere

disa

dvan

tage

sof

‘obj

ectiv

e’ a

ppro

ach

beco

me

obso

lete

.

Indi

vidu

als

know

the

spec

ific

requ

irem

ents

of

thei

r sp

ecifi

c jo

b be

st(s

peci

fical

ly: b

ette

r th

anla

bour

mar

ket e

xper

tsge

nera

ting

the

GE

Dsc

ale)

.

Dev

elop

ers

of a

p-pr

oach

and

ear

lyst

udie

s

Qui

nn a

nd M

andi

lovi

tch

(197

5),

Qui

nn a

nd S

tain

es(1

979(

a), (

b)),

Dun

can

and

Hof

fman

n(1

978,

198

1)

Sho

rt d

escr

iptio

n of

app

roac

h

IIa:

‘Sub

ject

ive’

app

roac

hE

ach

wor

ker

is a

sked

abo

ut th

e ed

ucat

iona

l req

uire

-m

ents

of h

is s

peci

fic jo

b (e

.g. ‘

Wha

t kin

d of

edu

ca-

tion

is u

sual

ly r

equi

red

to p

erfo

rm (

get)

a jo

b lik

eyo

ur’s

?’ (

requ

ired

educ

atio

n, (

RE

)).

The

sca

le o

f the

RE

var

iabl

e ha

s to

be

mad

e co

m-

patib

le w

ith th

e sc

ale

of th

e re

spon

dent

s in

form

atio

nof

acq

uire

d sc

hool

ing

(idea

lly a

lread

y by

the

desi

gn-

ers

of th

e qu

estio

nnai

re, e

. g. b

y as

king

for

year

s of

scho

olin

g re

quire

d).

Ove

rqua

lific

atio

n of

an

indi

vidu

al is

sta

ted,

if S

A >

RE

.


487

Box

: con

tinue

d

Sel

ecte

d fo

llow

-up

empi

rical

stu

dies

base

d on

app

roac

h

Mos

t of o

verq

ualif

icat

ion

stud

ies

auth

ored

or

coau

thor

ed b

y B

üche

l.

Dis

adva

ntag

es o

fap

proa

ch

App

roac

h pr

oduc

es a

so-c

alle

d ‘m

ixed

’ or

‘gre

y’ a

rea

of w

ork

with

doub

tful p

laus

ibili

ty o

fco

mbi

natio

n of

req

uire

dsc

hool

ing,

acq

uire

dsc

hool

ing

and

occu

pa-

tiona

l sta

tus.

In g

ener

al,

indi

vidu

als

wor

king

insu

ch a

situ

atio

n ha

ve to

be e

xclu

ded

from

over

qual

ifica

tion

anal

yses

. Thi

s le

ads

toin

form

atio

n lo

ss in

gene

ral a

nd r

educ

ednu

mbe

r of

cas

es in

spec

ial.

Num

ber

of m

issi

ngva

lues

in th

e pr

oduc

edov

erqu

alifi

catio

nva

riabl

e is

slig

htly

high

er th

an in

the

stan

dard

sub

ject

ive

two-

varia

ble

appr

oach

,be

caus

e th

ree

sour

ceva

riabl

es a

re in

volv

ed.

Adv

anta

ges

of a

p-pr

oach

Test

s sh

owed

a m

uch

high

er v

alid

ity w

hen

cate

goris

ing

over

qual

ifica

tion,

com

pare

d to

the

stan

dard

sub

ject

ive

two-

varia

ble

appr

oach

.

Dev

elop

ers

of a

p-pr

oach

and

ear

lyst

udie

s

Büc

hel a

nd W

eiß

huhn

(199

7(a)

).

Sho

rt d

escr

iptio

n of

app

roac

h

IIb:

Var

iant

of ‘

subj

ectiv

e’ a

ppro

ach:

thre

e va

riabl

es a

ppro

ach

Mea

surin

g ov

erqu

alifi

catio

n in

a fi

rst s

tep

as d

e-sc

ribed

in II

a.

Usi

ng a

3 th

ird v

aria

ble

(occ

upat

iona

l sta

tus)

for

valid

atio

n.

Val

idat

ion

chec

k is

don

e us

ing

a st

anda

rdis

edca

tego

risat

ion

syst

em. T

his

prod

uces

the

outp

utca

tego

ries

‘cle

arly

pla

usib

le c

ombi

natio

n of

the

thre

eva

riabl

es’,

‘cle

arly

impl

ausi

ble

com

bina

tion’

(in

Ger

man

GS

OE

P: a

bout

1%

of c

ases

), a

nd ‘d

oubt

ful

plau

sibi

lity

of c

ombi

natio

n’ (

in G

erm

an G

SO

EP

:ab

out 5

% o

f cas

es).

Ove

rqua

lific

atio

n of

an

indi

vidu

al in

the

case

of S

A >

RE

is s

tate

d on

ly, i

f the

val

idity

che

ck le

ads

to a

clea

rly p

laus

ible

res

ult (

othe

rwis

e: g

ener

atin

g a

mis

sing

val

ue).

Felix Büchel

488

is that categories of training requirementsmust be translated into equivalent years ofschooling. Although such translations havebeen done in the past, there is no consensuson the appropriate conversion’ (p.30). Of themany different measurement methods, eachof which has a perceptible direct influence onthe empirical results, three are presented byRumberger (1980, p.105), Kalleberg andSørensen (1973, p.221) and Burris (1983c,p.457) respectively. Furthermore, the DOT/GED system takes no account of the situationwithin a particular occupational category:47

‘The DOT measure of required schooling canclassify someone as overeducated when theperson has a very good job within an occupa-tion’ (McGoldrick and Robst, 1996, p.281).

The final criticism is that the GED system,which dates from the sixties, does not respondto changes in the requirements for specificoccupations, such as those resulting fromtechnological progress (cf. Clogg et al., 1986,p.382; for a complete summary of the validityand reliability problems of the DOT/GED ap-proach, see Rumberger, 1981b, pp.59 et seq.).

Given the objections enumerated above, someof which have serious implications, it is notsurprising that this approach is hardly everadopted any more in new studies on over-qualification, which are largely based on theso-called ‘subjective’ strategy described in thenext point.

4.1.2 The ‘subjective’ approach

The reservations about the DOT/GED ap-proach which were expressed in the previous

point relate to both of the steps that need tobe taken to establish the education require-ment for the practice of a particular occupa-tion, i.e. the assignment of GED scores to theoccupations in the DOT list and the calcula-tion of the GED equivalent in years of school-ing. This criticism is not levelled, however, atthe idea of equating the required number ofyears’ schooling with the actual duration ofan employee’s schooling. It follows from thisthat any improvement, especially if it relatesto the reliability of the mismatch indicator,should focus exclusively on the process of de-termining of the education level required forthe job

One effective way of tackling the problem isto ask employees themselves for a subjectiveassessment of the qualification level requiredfor their respective jobs. Respondents may beasked to choose from a scale of requirementcategories or to assess the qualification levelin terms of a number of years of training. Thequestions are sometimes varied. Respondentsmay be asked for the qualification that is nor-mally required for the work they perform (thisis the case in the German Socio-EconomicPanel (GSOEP) study and in other surveyssuch as the employment survey conducted bythe Federal Institute for Vocational Training(BIBB) and the Institute for EmploymentResearch (IAB) in 1991-92). Another commonquestion asks for the qualification requiredto obtain the relevant job; this is the approachadopted in the Panel Study of Income Dynam-ics (PSID), although McGoldrick and Robst(1996, p.281) rate the PSID question inferiorto that of the GSOEP. The nature of the ques-tion does not alter the process of equating therequired qualification with the employee’sactual qualification level in order to obtainthe mismatch indicator.

The first such survey of job-requirement lev-els was conducted in the PSID of 1976, andothers took place in 1978 and 1985. The sameapproach was adopted by the GSOEP.

Although there are no reports in the relevantliterature of reliability problems stemmingfrom incomprehension (cf. Hersch, 1991,p.141), the subjective nature of the questionundoubtedly poses its own reliability problem.

47 What is meant here is a highly generic indica-tion of the combination of status group and job-requirement level, along the lines of the indicatorberufliche Stellung (‘occupational status’) which isoften available in Germany, for example in the fol-lowing form: ‘employment as: unskilled worker,skilled worker, foreman/master craftsman, white-collar worker in an unskilled job, white-collarworker in a skilled job, white-collar worker in ahighly skilled job, sole trader, self-employed per-son with employees, assisting family member,trainee/internee, clerical-grade civil servant, ex-ecutive-grade civil servant and administrative-grade civil servant’.


489

It is conceivable, for instance, that reportedrequirement levels will tend to relate to con-ditions of recruitment at a particular timerather than present job descriptions; moreo-ver, it is not impossible that cognitive disso-nance might cause overqualified employeesto imagine a higher requirement level thanthat which actually obtains (Hartog andOosterbeek, 1988, p.186-187; on this aspectof data collection, see also Kalleberg andSørensen, 1973, p.236).

Nevertheless, it may be expected that differ-ences in required skill levels for people withequivalent qualifications will be sufficientlyidentifiable: ‘Although this measure may con-tain much noise, due to differences in stand-ards the individuals employ, it will certainlybring out differences among equally educatedindividuals in the demands that their jobs puton them (Hartog, 1985a, p.282; for a similarline of argument, see Witte and Kalleberg,1995, p.301). For all its recognised flaws, thesubjective approach is generally held to bemore effective than the DOT/GED system:‘This [i.e. the subjective approach] has theadvantage of obtaining information from thesource closest to the actual job situation, tak-ing account of all specific circumstances’(Hartog and Oosterbeek, 1988, p.186). A gen-eral reliability check on this strategy, involv-ing a comparison of employees’ and employ-ers’ responses, for example, has yet to bedeveloped.

4.1.3 Unorthodox measurementstrategies

Studies devoted to the examination of mis-matches between job and qualification arecharacterised by an almost unbounded diver-sity of measurement strategies. The purposeof the following paragraphs is to show justhow diverse these strategies can be by meansof some examples from German literature.

Apart from the studies by Rippe (1988),Schwarze (1993), Plicht et al. (1994), Brink-mann and Wiedemann (1995) and the author,German employment research gives the im-pression of having been insensitive until quiterecently to the methodological debate outlinedabove, which was conducted with great gusto

in the pages of the U.S. journals. The diver-sity of methods not only raises questionsabout the reliability and validity of the meas-urement strategies but also complicates thequest for comparability of empirical data.

One classic measurement strategy in the fieldof German qualification research seeks toidentify the extent to which knowledge andskills acquired in vocational training are us-able in an individual’s current job;48 similarapproaches are also to be found in earlier lit-erature from the English-speaking world (e.g.Staines and Quinn, 1979, p.9). The problemwith this strategy is that there are basicallyno jobs in which employees can draw uponall the knowledge and skills they had to as-similate in the course of their vocational train-ing (cf. Suda, 1979, p.153). The assessmentof the degree of applicability is likely to bevery highly subjective. A validity test byHalaby (1994) on this type of measure pro-duced disappointing results; Rippe (1988,pp.179-180) could only pour scorn on thismeasurement strategy.

Kaiser et al. (1980) used a system of sevenjob indicators designed to assess the job/quali-fication match – indicators such as the de-gree of variety in the job, cooperation withsuperiors, etc. The validity problems of thisapproach seem obvious (p.102).

Krämer (1982) establishes the degree of job/qualification match for executives of smallcompanies by comparing their formal qualifi-cations with a job grading that he assigns onthe basis of their decision-making powers. Onthe basis of Krämer’s findings, it is doubtfulwhether his measurement strategy can beconsidered a success. For example, he reportsthat ‘It can actually be seen from the tablesthat all university graduates [in executiveposts] are overqualified for the work assignedto them’ [!] (p.105).

48 See for example Stooss (1979), Chaberny et al.(1982), Minks and Filaretow (1993), Jansen (1993),Teichler (1994), Lewin et al. (1994a and b),Hofmann and Vogeler (1995) and Pfeiffer andBlechinger (1995).

Felix Büchel

490

Another approach tries to find a means ofidentifying overqualification on the sole ba-sis of occupational status and formal qualifi-cation (see for example Henninges, 1991;Minks, 1992; Tessaring, 1994a; iwd, 1994a;Minks, 1996, and Velling and Pfeiffer, 1997).This strategy is clearly too basic to obtainvalid results. The range of job requirementswithin various status groups is too wide toallow unequivocal categorisation. This is il-lustrated, for example, by the questionwhether or not a desk-officer post in an ad-ministrative body should be categorised asappropriate employment for a graduate of ahigher technical college.49

As an alternative to occupational status, useis also made of the employee’s position withina company, a category that is available fromsurveys such as the German microcensus (seefor example Tessaring, 1994a, and Velling andPfeiffer, 1997). A combination of the two sta-tus categories is used by Handl (1996), whouses information on both occupational statusand company position to establish the levelof an individual’s employment. This, however,is unlikely to solve the aforementioned valid-ity problems.

Earlier works sometimes had recourse to theemployee’s pay level as a suitable criterion ofadequate employment (see for exampleTessaring, 1984). The findings produced byRumberger (1980), however, testify to theproblems arising from income analyses in con-nection with the acquisition of data onoverqualification: during the period coveredby his study, the income position of collegegraduates in relation to high-school graduateshad not deteriorated, but the relative ad-equacy of their education had. Schlegelmilch(1982) applied a more refined strategy, com-bining reported data on occupational statusand income level, on the basis of which she

classified as underskilled any job done by agraduate for which no university degree wasrequired and for which the incumbent re-ceived less than the minimum rate of pay forgraduates in the German public service, i.e.point IIa on the statutory salary scale for fed-eral employees (see p.407, footnote 31).

An original approach is used by Szydlik, atleast in his studies in German (Szydlik, 1996a,1996b and 1997a). He extends the subjectiveapproach by combining the obtained variablesrelating to the job/qualification match withthe question whether the employees work inthe occupation for which they have beentrained. The question is what additional in-formation this disaggregation yields. In a ver-tical sense (and this refers not only to the basicprinciple underlying the strategy for meas-uring overeducation but also the line of argu-ment used by Szydlik in the interpretation ofhis findings), the information that a personis not working in the occupation for which heor she has been trained cannot be regardedper se as proof of a mismatch; in some cases,job descriptions in different occupations arefar too similar to permit such an assertion (cf.section2 of the present study).

A highly unconventional categorisation ofmismatches can be found in Wonneberger(1994, pp.144-145), where a mismatch isdeemed to exist if a person is trained for anoccupation (for which an academic degree isnot required) in which career prospects arepoor.

The diversity of measurement strategiesillustrated in the foregoing paragraphs un-derlines the soundness of Rumberger’s pos-tulation of the need for standardised measure-ment of overqualification (Rumberger, 1994,p.281). Teichler (1994, p.28) reports that vari-ous studies have put the percentage of gradu-ates in overqualification at levels rangingfrom 40% to under 5%, depending on the cri-teria applied. Elsewhere, Teichler (1998, p.98)tells of an approach that involved assigningthe ‘overqualified’ label to all universitygraduates who admitted to knowing non-graduates with similar professional duties totheir own (!). In such a case, the size of eachrespondent’s circle of acquaintances would

49 See for example Minks and Filaretow (1993,p.43), Gleiser (1996) and even Plicht et al. (1994);on the discussion of this question, see also Vellingand Pfeiffer (1997, p.14, footnote 17); on the dif-ferentiation between graduates of technical col-leges and universities, see Gleiser (1996, pp.36-37).


491

have a direct bearing on the findings of thestudy. It is patently obvious that such meas-urement strategies can yield little useful in-formation.

4.2 An innovative approach

As a refinement of the familiar measurementstrategies presented in subsection 4.1 above,the empirical part of this study (section 6) willintroduce a new process for identifyingoverqualification. It is based on the conven-tional subjective approach. The innovativeelement is the inclusion of a third indicatorto validate the overqualification variableswhich were initially obtained from the infor-mation on employees’ formal qualificationsand job-requirement levels. It has to be saidthat the disadvantages of this strategy are aslightly higher rate of missing values and acategory entitled ‘implausible combination ofthe three basic variables’. In addition, thereis also an optional ‘mixed’ category to covercases in which the information conveyed bythe three basic variables does not permit aclear distinction to be drawn between a job/training mismatch and a case of over-qualification. The crucial advantage of thistriple-variable strategy is that it draws a farsharper line between adequate work andoverqualification than the conventional dou-ble-variable approach (for details see point6.1.2 below).

5. Overqualification:a literature survey

The educational policymakers’ fear of produc-ing surplus qualifications for which there isno demand in the marketplace is presumablyas old as institutionalised education itself.

Tessaring (1980, pp.374-375) corroboratesthis, quoting from almost 300 years of Ger-man educational history. For example, in 1890Bismarck pronounced these words of warn-ing: ‘One of the principal ills of our higherschool system lies in the surfeit of academicschools and in the artful seduction to attendthe same which our establishments practise,so that we breed learned young men far be-yond our needs and beyond any possibility of

their procuring suitable positions. Our higherschools are attended by too many young per-sons who are destined neither by talent norby parental provenance for a learned profes-sion […]. The consequence is the excessivestudy of all academic disciplines and the edu-cation of a proletariat of scholars constitut-ing a danger to the state.’ [author’s italics].

Similar quotations from other countries havealso been recorded. The debate on the careerprospects of qualified individuals leaving theeducation system was given a new lease oflife in the sixties and seventies, when a rapidexpansion of higher education took place al-most simultaneously in the United States andsome European countries, driven by demo-graphic developments and political decisions.Education and employment researchers in theUnited States were the first to undertake sys-tematic study of the consequences of a glut ofhigher qualifications and thereby set theirseal on the scientific examination of the phe-nomenon of overeducation.

5.1 Origins of the discussion

5.1.1 The U.S. discussion from thesixties to the eighties

Treatment of the overqualification problem byAmerican employment researchers may beconsidered to have begun with Berg’s mono-graph of 1970. Under the catchy title TheGreat Training Robbery, Berg tried to dem-onstrate that the labour market was no longerable to absorb the output of university andcollege graduates comfortably following thesharp rise in their numbers in the wake ofthe baby boom and changes in educationalpreferences.50 The repercussions would bekeenly felt in the form of unemployment andoverqualification, accompanied by falling re-turns to education. As the Bureau of LaborStatistics later established, the number ofuniversity and college graduates in under-skilled jobs multiplied from about a millionto 3.6 million in the short period from 1969 to1980 (Hecker, 1992, p.5).

50 See also his subsequent works on this subject,such as Berg (1989).

Felix Büchel

492

Methodological innovation made a very earlyappearance in the study by Kalleberg andSørensen (1973), which presented a simplemeasurement strategy for the collection ofempirical data on overqualification. This ap-proach was based on the DOT/GED system(for details see subsection 4.1 above) and fol-lowed on from the work of Eckaus (1964) andScoville (1966), who took the measurementsprovided by the authorities for assessing therequired skill level for jobs in the various oc-cupational categories and adapted them foruse by employment researchers. Thesemultivariate data analyses did not yet con-trol, however, for significant factors such asage.

In 1975, following thematically related pre-liminary studies (Freeman, 1971, and Free-man and Breneman, 1974) one of the twomain protagonists in the U.S. overeducationdebate, Richard Freeman, entered the scene.His question Overinvestment in college train-ing? (Freeman, 1975b) is intended to be rhe-torical. Freeman was firmly convinced thatthe college system in the United States hadbeen producing a large surplus of qualifica-tions since the sixties, which was not entirelydue to the demographic effect of the babyboom. He tried to underpin this assertion insubsequent works, including a monographdevoted to the issue (Freeman, 1976b)51 anda contribution to the Review of Economics andStatistics (Freeman, 1977); this period alsosaw the appearance of his special studies de-voted to specific occupational groups (Free-man, 1975a, 1975c and 1976a).

The ‘discovery’ of the growing imbalance be-tween the supply of and the demand for higherqualifications in the U.S. labour market hadalready been made at that point (see for ex-ample Carnegie Commission on Higher Edu-cation, 1973), but it had not yet been suffi-ciently ‘marketed’. The pithy titles of hisstudies on the subject of overqualification,suggesting that the hypothesis underlying hisresearch is an incontrovertible truth, and hissometimes drastic speculations on the conse-quences of the growing qualification surplus

(e.g. ‘destabilizing political consequences’ –Freeman, 1976b, p.189) achieved their aimand made Freeman the most-quoted authorin American literature on the subject ofoverqualification, even though numerous pa-pers in a similar vein appeared at that time(see for example Jenkins, 1974; Rawlins andUlman, 1974; Dore, 1976; Jaffe and Froom-kin, 1978; Brinkmann, 1978; Suda, 1979;Denison, 1979, and a later work relating toCanada, namely Dooley, 1986).

Freeman’s presentation, which is quite one-sided at times, could not fail to provoke criti-cism (see for example Smith and Welch, 1978).Several replies were published in issue No1of the Journal of Human Resources for 1980.Witmer (1980) and Schwartz and Thornton(1980) criticised Freeman’s methodologicalapproach.

Russell Rumberger, who was to emerge as thesecond main protagonist in this discussionalongside Freeman, did not find any empiri-cal evidence to suggest that the relative posi-tion of university and college graduates in thelabour market was deteriorating, but did iden-tify a rise in overeducation (Rumberger, 1980).Freeman, however, was unmoved by thiswelter of criticism, describing the claims ofSchwartz and Thornton, for example, as ‘purenonsense’ (Freeman, 1980, p.141). In a con-cluding review of the debate in the Journalof Human Resources, however, Kaufman(1984) advances arguments in support ofSchwartz and Thornton. After the publicationof another two shorter papers (Freeman,1981a and b), Richard Freeman withdrewfrom this field of research.

Staines and Quinn (1979) adopted an inno-vative approach, analysing individual data onjob characteristics in a partial time series.Survey data for the years 1969, 1973 and 1977are examined. One category in the collecteddata is the respondents’ subjective assess-ment of their job/education match (p.9).Staines and Quinn also arrive at the conclu-sion that the overall degree of mismatch hastended to increase.

In subsequent studies, Rumberger (1981a,1981c and 1984) concentrated on presenting51 For a review of the monograph, see Levin, 1977.


493

evidence in support of his thesis that the de-velopment of the job structure in the UnitedStates is marked by a slower rise in skill re-quirements than would be necessary to en-sure that enough appropriately skilled jobswere available for the masses of new univer-sity and college graduates entering the labourmarket. He even asks the question whethertechnical progress, contrary to common be-lief (see for example Cappelli, 1993, who ex-amines the production system), actually leadsto a reduction in the skill level required forthe average job (Rumberger, 1981a, p.588; seealso Rumberger, 1981b, p.67).

Rumberger’s studies are innovative in rela-tion to Freeman’s in that Rumberger shiftsthe concept of overeducation from themacroeconomic level to that of the individual;all of his measurements are based on theDOT-GED approach.52 His monograph onovereducation in the U.S. labour market(Rumberger, 1981b) – the most important onthe subject of overqualification along withFreeman’s The Overeducated American andperhaps also T.A. Sullivan’s Marginal Work-ers, Marginal Jobs (Sullivan, 1978) – goes farbeyond any previous literature in its compre-hensive and thematically broad portrayal ofthe researched aspects of overeducation.

Another major innovation in the study ofovereducation is to be found in Duncan andHoffmann, 1981 (see also the Duncan andHoffmann study of 1978, which was the fore-runner of this work). These two authors ex-amined at an individual level the financialreturns to necessary, surplus and deficit com-ponents of education, thereby establishing adirect link to the human-capital approach. Inplace of the DOT/GED system, they used sub-jective data obtained directly from employeesabout the skill level required for their respec-tive jobs; this information had originally beencollected from the 1976 survey batch of thePanel Study of Income Dynamics (PSID). Asimilar approach underlies studies byRumberger (1987) and Shockey (1989).

New dimensions were added by Burris(1983a), whose study was the first to explorethe sociological and political aspects ofoverqualification. Until then, studies hadonly analysed individual sociological orpolitical aspects, such as the effects ofoverqualification on health (see for exampleKasl, 1974; House, 1974; Coburn, 1975, andCaplan et al., 1980).

The studies by Jaffe and Froomkin (1978),Clogg (1979), Clogg and Sullivan (1983), Cloggand Shockey (1984 and 1985), Clogg et al.(1986) and Lichter (1988) – although Lichter’sdefinition of overqualification poses someproblems – proved that the rise in over-education in the United States was caused toa great extent by changes in the demographicstructure of the potential labour force.53 Thestudy by Burris (1983b) can be considered aspart of this cluster, too.

The 1985 study by Tsang and Levin seeks topresent the first integral economic theory toexplain the persistence of overqualification;the study focuses primarily on productivityissues (for a critical appraisal, see deGrip,1989; for a reply to this criticism, see Tsangand Levin, 1989.

By 1986, the postulate that overeducation wasgradually developing into a critical problemwithin the U.S. labour market had come tobe regarded as an established fact, but it wassubjected to critical examination for the firsttime by Smith (1986); however, this methodo-logically based study, which was sharply criti-cal of the measurement strategies used insupport of the said postulate, had a limitedimpact.

5.1.2 The early discussion outside theUnited States: the case of Germany

Outside the United States too, the sometimesdramatic expansion of educational provisionwas the subject of controversy in various coun-tries (for a general review, see Teichler, 1996,

52 For a detailed description of the DOT/GED sys-tem, see point 4.1.1 above.

53 Dooley (1986), however, arrives at the oppositeconclusion for Canada.

Felix Büchel

494

pp.25 ff.). It must be said, however, that theacademic quality of the debates lagged farbehind those in which the U.S. researcherswere engaged, at least until the late seven-ties. The case of Germany may be cited as anexample.

The study Akademisches Proletariat?(Schlaffke, 1972) caused quite a stir amongGerman educational researchers. The study(in fact, the term ‘essay’ is more apt) strikesa perceptibly and no doubt deliberately po-lemical tone. The line of argument is inher-ently negative and fuels conservative preju-dices; moreover, the author generally failsto provide his readers with sources in whichthey might find empirical figures to substan-tiate his hypotheses. The following quote ex-emplifies this approach: ‘How can a politicalscientist whose special papers at universitywere in the theory of the class struggle andstrategies for overthrowing the system pos-sibly explain the workings of our economicsystem to his pupils in a modern-studiesclass?’ (p.47). If strict academic standards ofempiricism and analysis are applied, this keycontribution to the early German literatureon overqualification is clearly not in any re-spect in the same league as its American con-temporaries.

The only field of research which seemed tointerest German empiricists was that of edu-cational forecasting. As in the early discus-sion in the United States, it was not a matterof studying aggregated individual discrepan-cies between job requirements and formalqualifications but rather of alleged or expectedimbalances between the aggregate supply oflabour with particular qualifications and theaggregate demand for such labour. Besidesthe interest in quantitative forecasting, themain methodological bone of contention waswhether the demand-centred manpower re-quirement approach or the supply-centredsocial demand approach should be adopted(for a detailed review of the host of conflict-ing predictions, see Tessaring, 1980, pp. 376).The huge discrepancies between the findingsof such studies, however, presumably sowedmore confusion than reassurance in the mindsof educational planners (cf. Schlaffke, 1972,pp.10 et seq.).

Not until the end of the seventies did empiri-cal research in Germany diversify into sev-eral different subject areas. Tessaring (1978)tried to trace the careers of graduates but hadto rely on highly aggregated statistical mate-rial (see also Tessaring, 1981 and 1984, forsimilar approaches). A breakthrough wasundoubtedly made in the empirical study ofgraduate employment when the micro-economic Infratest surveys were conducted inthe autumn of 1978 and the spring of 1979(see Stooss, 1979). Similarly designed analy-ses by the Institute for Employment Research(IAB) around that time laid the foundationsfor a new research tradition (see for exampleKaiser et al., 1980).

The competition that emerged in the late sev-enties and early eighties between micro-economic empirical research, which strove toestablish certain methodological standards,and the ‘essayist’ approach to the over-qualification problem is exemplified by thepapers published in an anthology of contri-butions to the Conference on University Ex-pansion and the Labour Market – Problemsand Research Perspectives (Tagung «Hoch-schulexpansion und Arbeitsmarkt – Problem-stellungen und Forschungsperspektiven»),held in Berlin in November1981. While somesound empirical papers were presented(Tessaring, 1983, and Schlegelmilch, 1983a,for example) there was certainly no shortageof representatives of the ‘old school’. Buschand Hommerich (1983), for instance, saw thecause of the escalating employment problemin changing ‘power constellations’ (p.71),while Krais (1983) proposed that the issue ofgraduate employment and career prospectsbe examined ‘in the light of class theory’(p.36), and Nuthmann (1983) expressed thefollowing view on the graduate overqualifi-cation: ‘Some of these activities, work formsand lifestyles may be directly connected withlabour-market problems of the people con-cerned. In other cases, they are probably areaction to growing economism and industri-alisation, bureaucratisation and the increasein the statutory regulation and control of allaspects of life’ (p.11). In the years that fol-lowed, however, such approaches were almostentirely supplanted by empirical analyticalresearch (see point 5.2.2 below).


495

It only remains to describe how the GermanFederal Government assessed the threat ofan increase in sub-standard forms of employ-ment for graduates as a result of the reformand expansion of higher education. The Fed-eral Ministry of Education and Scienceseemed to be rather wary of vastly overopti-mistic predictions of labour demand. The pur-pose of a course of study, namely to qualifythe student for professional activity of a suit-able standard, certainly remained intact (‘Aimof the course of study: The purpose of instruc-tion and study shall be to prepare the stu-dent for a field of professional activity [… so]that he or she is capable of performing aca-demic […] work’ – section 7 of the Higher Edu-cation Framework Act (Hochschulrahmen-gesetz), quoted in Neusel, 1983, p.236).Expectations regarding the ability of the la-bour market to play its part in the achieve-ment of this aim, however, were muted. Aspart of the process of developing ‘points ofreference’ for the Permanent Commission onHigher Education Reform, the Federal Min-istry of Education and Science called on em-ployers to ‘create jobs [for graduates] belowthe top level’, and expressed confidence thateven ‘“underskilled” occupations’ could ‘be ameans of job satisfaction’ (quoted in Neusel,1983, p.240).

5.2 Present state of research

5.2.1 Recent studies relating to theUnited States and Canada

Recent studies from the United States on thesubject of overeducation/overqualificationhave been characterised by the fact that, un-like the studies referred to in subsection 5.1above, they rarely focus on the consequencesof a glut of university and college graduatesin the labour market. By the end of the eight-ies, the concept of ‘overeducation’ – relatingto individuals – had become established as aseparate category in the field of employmentresearch, and subsequent studies becamemore widely diversified.

For the first time, the implications ofoverqualification for business managementcame under scrutiny. The question waswhether overqualified employees were less

productive in the same level of job than theiradequately educated colleagues (Tsang, 1987,and Tsang et al., 1991). The relevant studieswas based on in-house data of the U.S. BellCompanies. More recent studies examine thecorrelation between overqualification andhealth (Amick and Lavis, 1998) and betweenoverqualification and job satisfaction (Hersch,1991, Johnson and Johnson, 1992, 1996 and1997, and Johnson and Roy, 1994 and 1995).

A latecomer to the debate described in sub-section 5.1 above was the article by Verdugoand Turner Verdugo (1989), which followedon from a shorter study (Verdugo and TurnerVerdugo, 1988). Adopting a similar approachto Duncan and Hoffmann (1981), but withoutexplicitly referring to that approach, the au-thors believed they could disprove the Duncanand Hoffmann findings, because they calcu-lated negative returns to components ofschooling that employees did not require inthe performance of their jobs. Their studyaroused vehement opposition (Cohn, 1992,Gill and Solberg, 1992, then Cohn and Khan,1995); the reply by Verdugo and TurnerVerdugo (1992) to the criticism of their line ofenquiry and of the way they interpreted theirfindings was rather unconvincing.

Sicherman (1991) studied overqualification inthe framework of the career mobility theory(Sicherman and Galor, 1990). He used thephenomenon of overqualification as a practi-cal example of the empirical relevance of thetheory he had helped to develop. He dismissedthe possibility that other theories could domuch, if anything at all, to explain the phe-nomenon. An extensive review (and endorse-ment) of the career-mobility approach is con-tained in Hersch (1995).

The observation that the imbalance in thelabour market for graduates which had be-gun to appear in the sixties had continuedthroughout the eighties (and that no improve-ment was in sight – cf. Shelley, 1992 and 1994)was the subject of another series of ‘conven-tional’ articles (Hecker, 1992, Tyler et al.,1995a, b and c) who criticised Hecker, and thelatter’s reply to this criticism (Hecker, 1995a);see also Amirault, 1992, Cappelli, 1993,Hecker, 1995b, and Zemsky, 1998.

Felix Büchel

496

An article by Halaby (1994) levelled methodo-logical criticism at the DOT/GED approachand at questions on the occupational utilityof knowledge acquired through formal edu-cation, to which the article ascribes little va-lidity.

The study by Bishop (1995) gives a short in-troduction to the topic of overeducation andthe respective state of research in form of acontribution to an encyclopedia of economicsof education. In the same volume, the articleby Levin (1995) presents a broader overlookto the environmental setting, in which thisspecial field of research is done.

The connection between the quality of a uni-versity or college education and the probabil-ity of overqualification is examined by Robst(1995b). As expected, Robst establishes anegative correlation between the quality ofthe university or college a person attends andthe probability of subsequent overqualifi-cation.

Lastly, an original study is presented byMcGoldrick and Robst (1996), who test thetheory of ‘differential overqualification’ devel-oped by Frank (1978b); the originality of theirapproach lies in the fact that, instead of bas-ing their test on income factors (see for ex-ample Ofek and Merrill, 1997), they use anemployee-centred measure of overquali-fication. They reject the hypothesis that mar-ried women run a higher risk of overqualifi-cation in smaller local labour markets.

Madamba and De Jong (1997) reopen the dis-cussion on the influence of demographicchanges on the degree of overqualification.They find that, in the United States, maleimmigrants from India, China and Korea aremore frequently overqualified, those from Vi-etnam are more rarely unemployed, whileimmigrants from the Philippines and Japando not differ significantly in this respect fromthe longer-established Anglo-Saxon popula-tion. Zhou (1993), on the other hand, reportedhigher mismatch rates for Japanese (andCuban) workers. To what extent the qualifi-cations of those immigrants who were edu-cated in their native countries can be madecomparable is an exceedingly moot question,

given the extreme variations in the quality ofthe school systems in their countries of ori-gin and is a fundamental problem in studiesdesigned to examine overqualification amongimmigrants.

On the basis of Canadian data, Vahey (forth-coming) certainly confirms in principle, atleast as far as men are concerned, the find-ings of Duncan and Hoffmann (1981) andtheir epigones, according to which positiveand negative returns to human capital maybe expected for surplus and deficit years ofschooling respectively; he demonstrates, how-ever, that the returns vary in accordance withthe skill level required for each job. Forwomen, on the other hand, he finds insignifi-cant effects at all job levels.

5.2.2 Recent studies relating to thecountries of the European Union

The main distinct traditions of research intooverqualification outside the United Statesexist in the Netherlands and Germany. In theUnited Kingdom the subject was not ‘discov-ered’ until the mid-nineties, although the sub-sequent volume of research on overquali-fication in the UK has been considerable.There are only isolated analyses for othercountries.

The Netherlands

There are several analyses of job/qualificationmatches in the Netherlands. If we focus ex-clusively on the studies devoted to overqualifi-cation, it emerges that Dutch employmentresearchers, compared with their colleaguesin other EU countries, have been the mostprolific contributors on this subject in Eng-lish, at least in the widely accessible English-language publications.

The first studies were undertaken by Hartog(1985a and 1986). His approach is very simi-lar to that of Duncan and Hoffmann (1981),although he does not cite the latter work. LikeDuncan and Hoffmann, he rejects unilateraldetermination of pay rates for the Dutch con-text too, either by the supply side throughformal qualification (the human-capitaltheory) or by the demand side through job


497

requirements (the job-competition model). Inthe good old Tinbergenian tradition, he pro-vides evidence of the explanatory capacity ofthe assignment theory. Extending this ap-proach, Hartog and Oosterbeek (1988) iden-tify a gradual rise in overqualification in theNetherlands over a period of time. Groot(1993a) refines the approach presented inDuncan and Hoffmann (1981), devoting spe-cial attention to the link between trainingactivities and mismatches. Overqualification,Groot says, not only has an adverse effect ontotal income from employment but also hasthe same sort of impact on returns to in-serv-ice training.

In more recent papers on the subject, Hartog(1997 and 1999b) passes critical judgementon the body of research into overqualification.As far as future research perspectives are con-cerned, the author puts the case for longitu-dinal analyses, more theoretical input, moreambitious evaluation procedures and – in theold-established tradition of Dutch research inthis field, which bears the unmistakable im-print of Tinbergen – for analytical approachesin which supply and demand effects are con-sidered in combination. In the study byHartog and Jonker (1998), the authors haveaccess to the Brabant-survey which includesIQ measures. Having the opportunity to con-trol for individual ability opens the door tonew aspects. The authors, however, find thatthe impact of this variable on the risk to workovereducated is relatively low. In other stud-ies (Hartog, 1999c, d and a), Hartog once againwarns against investing overqualificationwith unduly pejorative connotations, referringto the positive, albeit limited, returns to sur-plus years of education which were previouslyidentified by Duncan and Hoffmann (1981).

Borghans and Smits (1997) show that an in-creasing percentage of overqualified labourcan also impair the earning potential of ad-equately educated employees. DeWitte andSteijn (1998) examine the effects of increas-ing automation on the incidence of overquali-fication. They discover a direct link as wellas frustration effects in overqualified employ-ees. Batenburg and Witte (1998) identify arise in the mismatch rate within the Dutchlabour market over the period from 1977 to

1995. Groot and Maassen van den Brink(1999) demonstrate the sensitivity of themeasurement strategies used to identify jobswhich are incommensurate with their incum-bents’ qualifications. Groot and Maassen vanden Brink (forthcoming) present a survey ofempirical findings from various countries.Among other things, this survey reveals apositive correlation between growth rates andoverqualification rates. Van Eijs and Heijke(forthcoming) demonstrate that the currentallocation of qualifications to job requirementsis inefficient; this finding contrasts withHartog’s latest estimates. The study byBorghans and de Grip (1999) examines de-terminants of overqualification, categorisingthem by whether or not they are compatiblewith the allocation theory. A clear result, how-ever, does not emerge. Van der Velden and vanSmoorenburg (1999) compare results gainedfrom the so-called objective and subjectivemeasurement approach. Their major findingis that overqualification measured with anobjective approach is clearly overestimated.On the other hand, they do not find hints foran underestimation of overqualification whenapplying the subjective approach.

Two Dutch authors are about to present a newanthology on overqualification (Borghans andde Grip, forthcoming(b)). In their editorial ar-ticle (forthcoming(a)), Borghans and de Gripdiscuss the current issues in this field. Thepaper by Muysken and terWeel (forthcoming)seeks to integrate human-capital theory, job-search theory, the job-competition model andscreening into a discrete theoretical approach.Borghans et al. (forthcoming) examine theconnection between aggregated and indi-vidual overqualification on the one hand andlow pay on the other and conclude that theimpact of overqualification is greater thanthat of low pay. Finally, the study by Gautier(forthcoming) examines cyclical effects on thelevel of demand for various qualifications. Oneof his findings is that, during recessions, em-ployees in unskilled jobs are paid off withoutregard to their qualification levels.

Besides the aforementioned works, there areseveral relevant essays in Dutch, such asHartog, 1985b, Oosterbeek, 1986, Groot,1993b, Groot and Maassen van den Brink,

Felix Büchel

498

1996, Oosterbeek and Webbink, 1996, andGroeneveld, 1997, which testify, as in the caseof Germany, to a wealth of research on thesubject of overqualification that the interna-tional scientific community is unable to ap-preciate. The author’s lack of knowledge ofthe Dutch language prohibits an evaluationof the content of these papers at the presentjuncture.

Germany

The only comprehensive monograph onoverqualification in Germany is a study bythe present author (Büchel, 1998b). One ofthe main innovative methodological elementsof this study is that it is the first to adapt theexisting instruments of dynamic unemploy-ment research systematically for use in theanalysis of overqualification. In this way, onthe basis of panel data, it is possible to ana-lyse the probability of individuals being re-cruited and of their staying with or leavingtheir firm, to determine longer-term incomeeffects and more besides. Important prelimi-nary work, especially with regard to the cat-egorisation of overqualification, was per-formed in the studies by Büchel andWeißhuhn (1997a and 1998). Schlegelmilch(1987) presented a monograph analysing dataon overqualified graduates; another mono-graph, in the form of a company study, on theeffects of overqualification among techniciansand engineers (Haugrund, 1990) was designedmore as a business-management resource.

Two individual papers dealing with the sub-ject under examination were produced bySzydlik (1996a and 1997a). To generate themismatch variables, Szydlik includes a ques-tion about the occupation for which the em-ployee has trained (for a critique of this ap-proach, see section 2 above) but does not useoccupational status to validate the mismatch(Szydlik, 1996b, p.300, footnote 10) as Bücheland Weißhuhn do (see also subsection 6.1 be-low). This produces distinctly higher percent-ages of overqualification than would resultfrom an operationalisation based on theBüchel/Weißhuhn approach. Velling andPfeiffer (1997) also examine overqualification(as part of a wider-ranging study) for all quali-fication levels, though for western Germany

only. Despite the use of a different databaseand a sharply divergent means of assessingthe degree of job/qualification match, thestudy arrives at very similar findings to thoseof Büchel and Weißhuhn (1997a and 1998).Lastly, Zwick (2000 and forthcoming) adoptsan ambitious theoretical approach to exam-ine the links between overqualification, wagelevels and attitudes to education.

In continuation of the research tradition out-lined in point 5.1.2 above, the main focus ofGerman overqualification research has re-mained on university graduates, right downto the present day.

Straddling the border between ‘old’ and ‘new’research is the work of Rippe (1988), who wasthe first researcher in Germany to conduct abroad investigation of the methodological sideof job-match assessment and thus to reflectat least some signs of affinity with the bodyof U.S. research on that subject; until thatpoint, Germans had been virtually obliviousto the discussion of these issues in the UnitedStates. Proceeding from the DOT/GED ap-proach and the criticism levelled at it and atits application to the microcensus by IAB re-searchers, Rippe identified ‘a lack of validatedmethods for assessing job matches by com-paring existing and required qualifications’.

Plicht et al. (1994) presented the first broad-based study of the suitability of jobs held byuniversity graduates in Germany. Their ap-proach may not correspond exactly to theDOT/GED strategy, but it is related to thelatter in so far as it does not have recoursesubjective assessment of job-requirement lev-els by the respondents themselves. Althoughit is designed as a cohort study without longi-tudinal examination of individuals, the au-thoresses postulate that the high percentageof young graduates in jobs for which they areoverqualified is primarily due to the fact thatoverqualification is a natural phenomenon atthis stage in a graduate’s career. This inter-pretation was put into perspective in Büchel(1996a): although the risk of involuntaryoverqualification is higher at the start of acareer than after a period of occupational ex-perience, it has been proved that structuraland cohort effects have played a dominant role


499

during the nineties. In other words, the per-centage of graduates who tend to run a higherrisk of overqualification on account of theirmembership of particular status groups (e.g.women and graduates of technical colleges)is rising steadily, and membership of thesegroups is largely unalterable in the course ofa career.

The situation of academically trained careerstarters is examined by Büchel and Matiaske(1996) on the basis of longitudinal data fromthe German Socio-Economic Panel (GSOEP).It emerges from this study that a high risk ofoverqualification at the start of a career at-taches to certain ‘soft’ academic disciplineswhich are mainly chosen by female students.An interaction analysis shows that the gen-der effect is governed by the student’s choiceof subject.

Most of the other literature relating to themore recent discussion of the phenomenon ofgraduate overqualification in Germany maybe divided into two categories: the first com-prises highly specific surveys of graduates,differentiated by subject specialisation (mostof these have been conducted by the Univer-sity and College Information System (HIS) inHanover) or even by specific departments ofindividual universities (this type of survey isconducted by ‘interested parties’, i.e. studentsin one of the departments in question, oftenas one of their degree papers); the other cat-egory comprises papers presented at confer-ences on the subject of graduate career pros-pects and subsequently published in thereport of proceedings.

The HIS studies are reports on graduates inspecific federal states or subjects, based onthe system’s own surveys (see for exampleMinks, 1992 and 1996; Minks and Filaretow,1993; Lewin et al., 1994a and b). One of thecentral problems with these studies, however,lies in the database, which only covers thefirst years of a graduate’s career.

This problem, of course, also arises for uni-versity- and subject-related studies of gradu-ate employment. The utility of this type ofstudy is undisputed, because it can give pro-spective students important information

about future career prospects. Nevertheless,the scientific quality of the studies is gener-ally below average, since they are compiledas a rule by people with no research back-ground.54

The important thing about the latest devel-opments in this field is that the discussionabout the future of graduate employment inGermany now focuses on overqualificationand unemployment as closely related phe-nomena (see for example Schreyer, 1999, andWissenschaftsrat, 1999). This represents amajor step away from the narrow perspectiveswhich had long prevailed, in which unemploy-ment was only seen as an employment risk.

Studies on the overqualification of people withvocational diplomas are considerably lesscommon than studies on graduate jobmatches. The most plausible explanation ofthis phenomenon may be that higher educa-tion is expanding more rapidly than voca-tional education. Accordingly, graduates maybe regarded as more likely victims of labour-market imbalances.

Hofbauer and Nagel (1987) conducted a wide-ranging study of the risk facing skilled work-ers of occupational demotion to semi-skilledor unskilled status, examining the specificlevel of risk in various different trades, forexample. The study by Neubäumer (1993), towhich reference has already been made insection3 above, is based on an unconventionalapproach. The authoress proceeds on the as-sumption that segmentation applies not onlyto the labour market but also to the marketfor training places in the dual system of class-room and on-the-job training. The study dem-onstrates that in those occupations whereworking conditions are poor, training is pro-vided in excess of requirements as a means ofcoping with the loss of the better trainees, whomove upmarket when they have served theirapprenticeship. Those who complete theirtraining in occupations with poor workingconditions but do not find a job, either in theoccupation for which they have trained or in

54 For a collection of graduate studies from vari-ous European countries, see Teichler (1988).

Felix Büchel

500

a market segment with better career pros-pects (assuming they are prepared to switchto another occupation) are generally com-pelled to settle for overqualification. Neu-bäumer’s findings show that this phenomenonis undoubtedly empirically significant in Ger-many (see also Neubäumer (1997) andNeubäumer (forthcoming)).

A study by the present author (Büchel, 1994b)examines overqualification at the start of non-graduates’ careers. What seemed astonishingat first sight was the higher incomes earnedby the overqualified workers. This finding,which was also obtained by Hofbauer andNagel (1987, pp.54-55), shows that monetaryconsiderations are an important factor in theacceptance of an underskilled job, at least inthis phase of a person’s career (for a similarfinding, see Neubäumer, 1993). Pay differen-tials to compensate for monotonous work andother disagreeable working conditions, suchas the presence of noxious substances, couldalso play a part (cf. Lucas, 1997, p.557). Thisfactor may be particularly significant in thecase of trained craftsmen who subsequentlytake up semi-skilled jobs in manufacturingindustry or do piecework. Even after a shorttime, however, those who started out with agood fit would be better placed, because theirincome would have risen more sharply. A simi-lar approach is found in partial analyses suchas Büchel and Weißhuhn (1995), Weißhuhnand Büchel (1998), or Büchel and Pannenberg(1997).

Pfeiffer and Blechinger (1995) and Blechingerand Pfeiffer (1999) examine the realisablevalue of vocational training in the labourmarket and observe diminishing returns inthe course of people’s careers.

The only general study relating to Germanythat is exclusively devoted to the fit betweenvocational training and jobs was conducted,ironically enough, by two researchers from theUnited States (Witte and Kalleberg, 1995).The way in which they assess job/trainingmatches, however, is someewhat question-able, for they assert (p.311) that only about50% of men and 60% of women who have suc-cessfully completed their vocational trainingare in jobs that match their training [!].

An innovative study was compiled by vonHenninges (1996), who investigated the per-sistent contention that technical progress, theglobalisation of labour markets and other fac-tors would lead to the gradual disappearanceof unskilled jobs. In so doing, he addressedthe methodological problem whereby, becauseof the restricted availability of data, moststudies do not focus on the requirement lev-els for specific jobs but on the formal qualifi-cations of the people in those jobs. As a resultof the expansion of educational provision,there are indeed fewer and fewer workerswith no formal qualifications; however, thefact that unskilled work is being performedby semi-skilled or skilled workers means thatthe rate at which unskilled jobs are being shedis being systematically overstated (p.78).55

Lastly, the topic of overeducation was chosenas an example how to work efficiently withrobust stochastic earnings frontiers (Jensen,1999). This paper, however, has a strongereconometric focus than a substantial one.

Only a short time after the dissolution of theGerman Democratic Republic, special stud-ies on eastern Germany began to appear. VonHenninges (1991) shows that, contrary to theideological claims made by the regime,overqualification had already assumed epi-demic proportions during the Communist era(see also Schwarze, 1993, Büchel, 1995, andBrinkmann and Wiedemann, 1995, p.330).Schwarze (1993) presents a comparison of thesituation in eastern Germany before and af-ter the demise of the East German state (1989and 1991) and identifies income effects ofoverqualification.

The labour-market monitor for the new fed-eral states proved to be an effective source ofdata for the investigation of overqualification.In thematically wide-ranging IAB studies onthe development of the employment situationin eastern Germany, the phenomenon of

55 At this point, however, it should be mentionedthat, irrespective of data restrictions, macro-economic surveys of the jobs that people do are farmore difficult to conduct from a methodologicalpoint of view than surveys of employees’ qualifica-tions or occupational categories.


501

overqualification is given special treatment(Bielenski et al., 1994 and 1995, and Brink-mann and Wiedemann, 1995 and 1997). Thevery structure of the 1995 study by Brinkmannand Wiedemann (p.323) testifies to the highpriority that is attached to the phenomenon ofoverqualification in the empirical research ofthe IAB, at least in relation to eastern Ger-many, where the percentages of overqualifiedemployees are very high; this emerges espe-cially clearly from the special study on thissubject by Brinkmann and Wiedemann (1997).

These, then, are the studies that focus spe-cifically on eastern Germany.56

The extent to which foreigners in Germany areaffected by overqualification is investigated ina study by Szydlik (1996a). However, the factthat the source data (from the GSOEP) onlyyield rudimentary information on foreigners’qualification levels (at least for those who wereeducated in their native country) must poseproblems when it comes to assessing whethertheir jobs are commensurate with their train-ing. Szydlik observes, for example, that morethan half of the Turks working in Germanyare overqualified for their jobs. If we bear inmind, however, that even today Turkey re-quires only five years’ compulsory schooling ofits young citizens and that there is effectivelyno institutional system of vocational trainingin the country, the validity of this finding ap-pears dubious. Heinelt and Lohmann (1992)also elicit high overqualification rates for eth-nic Germans who have moved to the FederalRepublic from Eastern Europe; here too, thereare problems with the validity of the measure-ment strategy.

The job/training fit for working women is ex-amined in detail in one of the studies by Bücheland Weißhuhn (1997c). Extending the basic

approach adopted in Büchel and Weißhuhn(1997a and 1998), the authors attach specialimportance to an examination of the socioeco-nomic background of the women who are af-fected by overqualification. This relates notonly to the women’s family circumstances (e.g.the number of children in the household andthe arrangements made for their care) but alsoto their partner’s employment status, his in-come situation, etc. The study establishes astrong link between the financial circum-stances of the household and the need to ac-cept underskilled work, although marked dif-ferences were observed between women withdifferent qualification levels. The householdstructure and the number of children seem tobe less significant factors, with the qualifica-tion that women who are lone parents exhibita far higher risk of having to accept a second-best solution in the labour market.

Other countries of the European Union

In the United Kingdom, Sloane et al. (forth-coming – see also Sloane et al., 1996) werethe first to devote a study to the subject ofoverqualification. Their work is based on theresearch approach adopted by Duncan andHoffmann (1981) and delivers similarly struc-tured findings.

The Dutch researcher Groot (1996) found thatfew British employees were overqualified fortheir jobs. His assessment of the job/educa-tion match, however, was based on what ap-pears to be a somewhat arbitrary modifica-tion of the DOT system. An extension of thisapproach, based on the same assessmentstrategy, is found in Groot and Maassen vanden Brink (1995).

Battu et al. (1998a and b) test the theory ofdifferential overqualification (Frank, 1978b)and arrive at a negative result. In a longitu-dinal study, Battu et al. (1999) show thatoverqualification has adverse effects on jobsatisfaction and income. They produce thenoteworthy finding that social backgrounddoes not affect the probability of a mismatch.57

56 There are, however, other studies on the empiri-cal situation in eastern Germany, such as thoseby Büchel and Weißhuhn (1997a, 1997c and 1998)and by Szydlik (1997a, b and c). One remarkablepoint is the finding presented by Szydlik alone thatunderemployment is less widespread in easternthan in western Germany (see Szydlik, 1997b,Table 1, p.43, or the redundant findings in Szydlik,1997c, Table 2, p.13).

57 Cf. Patrinos (1995 and 1997), who arrives at theopposite result for Greece.

Felix Büchel

502

Alpin et al. (1998) examine over- and under-qualification in a wide-ranging study. Theirambitious quest to establish whether theoverqualification of individuals is only a tem-porary phenomenon meets with little success,however, because they only have access toright-censored data on job duration. Thisquestion cannot be answered on the basis ofsuch data. Finally, Battu and Sloane (forth-coming) find that displacement effects causedby overqualification do occur in the upper ech-elons of the labour market, but that unskilledworkers in the United Kingdom are not be-ing ousted to any great extent by overqualifiedrecruits.

One of the more unusual longitudinal stud-ies in the domain of overqualification researchwas presented by Dolton and Vignoles (1997).They observe that most employees who starttheir career in a job for which they areoverqualified do not manage to make the tran-sition to appropriate employment in the firstsix years. Dolton and Vignoles (forthcoming)arrive at similar findings to Sloane et al. (1996and forthcoming) with regard to the returnsto surplus years of schooling.

Alongside these microeconomic studies onoverqualification there are also conventionalstudies which examine the aggregate effectproduced by the expansion of the educationsystem, such as the study by Mason (1996).

For France there are studies by Forgeot andGautié (1997a and b), in which the percent-ages of over- and underqualified employeesin the younger age brackets are assessed byqualification level and job duration for theyears 1986 and 1995, as well as a study byVincens (1995), which treats the problem ofoverqualification in a rather general fashion(pp. 149-150).

With regard to Spain, there are relevant stud-ies by Alba-Ramírez (1993) and Beneito et al.(forthcoming). Alba-Ramírez finds evidence forthe job-matching and career mobility theories.Beneito et al. examine the question whethersurplus years of schooling should be regardedas complementary components of human capi-tal or substitutes; their test shows that thelatter is the more accurate assessment.

The Portuguese situation has been the sub-ject of studies by Kiker et al. (1997) and byMendes de Oliveira et al. (forthcoming). Kikeret al. adopt a similar approach to Duncan andHoffmann (1981) and find that it confirms theallocation theory. Mendes de Oliveira et al.refine the same Duncan and Hoffmann ap-proach and demonstrate that the returns tosurplus and deficit years of schooling areheavily dependent on job duration.

In Greece, studies on overqualification havebeen produced by Patrinos (1995 and 1997).He arrives at similar overqualification ratesfor Greek graduates to those recorded in west-ern Germany, for example. As might be ex-pected, Patrinos finds evidence of wide diver-gences between academic disciplines. Oneespecially noteworthy finding is that over-qualification is disproportionately high amonggraduates from humbler backgrounds.58

Other studies are based on the idea ofovereducation as an imbalance between thesupply of higher qualifications and the de-mand for them, an idea that was widespreadin the early days of the discussion (see sub-section 5.1 above); these include works byTsoucalas (1981), Psacharopoulos (1988),Glytsos (1990) and Lambropoulos and Psacha-ropoulos (1992).

For Austria there is a study on access to blue-collar occupations for apprentices when theycomplete their training (Ofner, 1994). Ofnerfinds that, two years after obtaining their cer-tificate of apprenticeship, about one-third ofthe former apprentices who have jobs areoverqualified.

This review of nationally focused literatureon the subject of overqualification shows thatthis major problem has yet to arouse the in-terest of employment researchers in manycountries of the European Union. We can butspeculate on the reasons for this. The re-stricted availability of key data may be a sig-

58 This finding is consistent with that of the au-thor (Büchel, 1997), who identifies a positive cor-relation between income prospects and the educa-tion level of the parental household, thereby chal-lenging the oft-expressed view that a universitydegrees is a social leveller.


503

nificant factor, for a mismatch variable can-not be generated without an adequate rangeof relevant reliable data. Efforts must be madeto ensure that data-gathering institutionsshow greater willingness than in the past toelicit the information that researchers need,such as the skill level required for individualjobs. This problem has already been recog-nised (ILO, 1998, p.90); a fair degree of pa-tience is likely to be needed, however, beforethis aim is achieved.

5.2.3 International comparative studies

There are studies in which findings from vari-ous countries are put together and some inwhich the researchers’ own empirical com-parisons are drawn.

Examples of the first group are the publica-tions by Suda (1979),59 Teichler (1988),Hofmann (1995) and more specifically thoseof Hartog (1997), Büchel and Weißhuhn(1997c), Borghans and de Grip (1999), Tessa-ring (1998), Büchel (1998b), and Groot andMaassen van den Brink (forthcoming). Thesestudies provide quite general information onvarious research findings in an internationalcontext.

Studies in which direct international compari-sons are drawn in the domain of overqualifi-cation have so far been limited to compari-sons between the country in which theresearchers are interested and the UnitedStates, which serves as the reference coun-try. Almost all such studies compare Germanywith the United States.

Büchtemann et al. (1993) presented a compara-tive study on career starts in Germany andthe United States. Their study analyses datafrom the PSID and GSOEP. Matches and mis-matches are only one aspect of this wider-rang-ing study. The study by Daly et al. (1997, re-vised in Daly et al. (forthcoming)) uses thesame databases to examine the income effects

of deficit and surplus components of humancapital along with the returns to qualificationswhich are actually needed for a person’s job.In so doing, it follows the approach establishedby Duncan and Hoffmann (1981). The authorsare interested in ascertaining whether struc-tural changes in the U.S. labour market havea greater impact on results than structuraldifferences between the countries under exami-nation. Astonishingly close analogies emergein the findings for Germany and the UnitedStates; the key finding of the study is that simi-larities between the nations are more pro-nounced than the similarities between the U.S.findings for different points in the time series.

Büchel and Witte (1997) analyse the effectsof overqualification in the initial phase of em-ployees’ careers. They use the longitudinalstructure of the ‘High-School & Beyond’ datasets and the GSOEP. Here too, the two coun-tries under scrutiny are remarkably similarin structural terms. The studies by Szydlik(1997b and c and, to a certain extent, a), arealso based on information from the PSID andGSOEP. While the aforementioned studies ar-rive at similar overqualification rates in theUnited States, the identified percentages ofunderqualified workers are strikingly lowerthan those calculated by Daly et al. (1997, Ta-ble 1, p.27).

Szydlik (1998) examines the overqualificationof immigrants in Germany and the UnitedStates. But this once again begs the questionhow, given the extremely wide divergence be-tween national training systems in terms ofquality, anyone can hope to achieve compara-bility between the formal vocational qualifi-cations that immigrants have obtained intheir own countries. Witte (1999) examinesthe development of overqualification amongyoung people in Germany and the UnitedStates; unlike Büchel and Witte (1997), thisstudy bases its analysis of the U.S. situationon PSID data. The study by Büchel andWeißhuhn (1997b) provides a summary inGerman of the main elements contained inthe previous studies by Daly et al. (1997) andBüchel and Witte (1997).

Cohn and Ng (1999) as well as the more fullyelaborated version of this study by Cohn et

59 The Suda study is listed here with certain reser-vations: while the theoretical part of this first studydeals explicitly with overqualification, the findingspresented in the empirical part are strangely un-related to the theoretical part.

Felix Büchel

504

al. (1999) examine overqualification in a com-parison between the United States and HongKong. What seems at first sight to be a ratherexotic research project actually produces re-markable results. In fact, the standard pat-tern that is familiar to us from Western re-search literature also applies to Hong Kong:as a rule, overqualification is more frequentlyobserved than underqualification, older em-ployees are more often underqualified fortheir jobs, while younger employees are morelikely to be overqualified, and the Duncan andHoffmann findings on the returns to surplusand deficit years of schooling, which were firstpublished in 1981 and have since been con-firmed in numerous studies, are likewisefound to hold true for Hong Kong.

The presentation of this literature survey hasalso shown up major research gaps. An im-portant research perspective for the futurerelates to the need to achieve mutually com-patible national data sets on mismatches andto analyse them on the basis of a standard-ised examination model. Only in this way caninstitutional and regional effects be satisfac-torily isolated; a comparison between indi-vidual studies for various countries cannotachieve this because of the widely divergentmeasurement methods.

6. Overqualificationand unemployment:The case of Germany

The purpose of this empirical section is topresent some examples, with the aid of a fewindicators, of links that exist betweenoverqualified and unemployed status.

In subsection1.2 above, a methodological af-finity was highlighted between overqualifiedand unemployed status within the labourmarket. The essence of this affinity is thatpeople in both situations are unable to capi-talise fully on their certified vocational quali-fications in the labour market. The ‘skill spill’in the case of overqualified workers amountsto x%, where x is only indirectly determina-ble – by means of an income analysis, for ex-ample – but is substantially higher than zero.

In the case of jobseekers, the spillage factoris 100%. If we assume, for the sake of sim-plicity, that the skill spill in the case of anadequately employed person is 0%, the ‘mid-dle’ ground of overqualification which liesbetween adequate employment and unem-ployment must be observable within thosesocial categories that unemployment litera-ture identifies as clearly disadvantaged.

6.1 Research approach

6.1.1 Database

The measurement strategy introduced in thefollowing paragraphs is based on data fromthe German Socio-Economic Panel (GSOEP)administered by the German Institute forEconomic Research (Deutsches Institut fürWirtschaftsforschung – DIW) in Berlin. Thisset of data, collected from representative sam-ples of the population resident in the FederalRepublic of Germany, contains a wealth oflongitudinal information on households andindividuals. The initial sample in 1984 com-prised almost 6000 households. The head ofeach household was given a questionnairecovering the whole household. All membersof the household aged 16 or over (more than12000 individuals) were also given a separatepersonal questionnaire seeking factual infor-mation and opinions on many aspects of theirlives, especially education and employment.

The households and individuals in this sam-ple are surveyed at yearly intervals. The da-tabase is constantly being widened as house-hold members leave to start new households,as new members move into panel householdsand as younger members of the householdreach the survey year in which their 17th

birthday falls.60

Information on the education and trainingof foreigners

The vast majority of the foreign nationals whoare GSOEP respondents are first-generationforeigners, in other words the ‘guest workers’

60 For more extensive information, see Projekt-gruppe Panel, 1995.


505

who were encouraged to come and work inGermany. In the present context, these peo-ple form a special group in the sense that theyunderwent schooling and vocational trainingin their native countries. There is only rudi-mentary information available on their edu-cational and training qualifications. This in-formation is not comparable with the veryprecise details elicited by the questions thatare addressed to German respondents abouttheir education level. Moreover, the educationsystems of the various countries of origin areso highly disparate that identical responsepatterns can rarely be taken to mean equiva-lent levels of qualification

The conclusion we must therefore draw fromthis situation is that it is impossible to meas-ure the degree of mismatch for foreign work-ers if they have been trained in their nativecountries. The aforementioned heterogeneityof training courses in their countries of ori-gin is not the only reason; another factor isthat the GSOEP question equates the vari-ous levels of job requirement with Germaneducational certificates. No very convincingattempts have yet been made to solve thiscompatibility and validity problem adequately(Szydlik 1996a);61 Szydlik’s exorbitant assess-ment that around 60% of all ‘qualified’ for-eigners are overqualified will certainly raiseeyebrows at the very least (Szydlik, 1996a,p.667; see also Szydlik, 1997a, p.17).

In the light of these considerations, the cat-egorisation of job/education matches in thefollowing paragraphs will only apply to thoseforeigners who have undergone at least theirvocational training in Germany. The problemsoutlined above with regard to information onimmigrants’ education levels also apply toother databases, but overqualification re-searchers have yet to respond to them withthe necessary sensitivity (see the criticism in

point 5.2.1 above of the methodology appliedto immigrant studies).

General case selection

All members of the sample who were in thelabour force at the time of observation arecategorised. People in full-time employment,regular part-time employment, minimal em-ployment, and involuntary unemployment(including the hidden labour reserve) are ex-amined, except in cases where the employeesare still in full-time education, since there willbe no ‘definitive’ information on vocationalqualification in such cases. For the same rea-son, internees, voluntary workers and peoplein similar occupational situations are ex-cluded from the analysis, as are those of re-tirement age (65 and over). Since people whohave no vocational qualifications at all can-not by definition be overqualified, they too areexcluded from the categorisation (cf. Witteand Kalleberg, 1995, p. 308, for an analogousapproach). With respect to the latter selection,an exception is made for the step of analysisresulting in Table 1.

The considerations described above form thebasis of the condition that only people whohave received vocational training and/orhigher education in Germany are taken intoaccount. Migrants who moved between EastGermany and the Federal Republic are alsoexcluded from the analysis.62 Because of thecategorisation rules governing the assessmentof overqualification (see point 6.1.2 below), itis also necessary to exclude those few peoplefor whom no information on vocational quali-fication, job-requirement level or occupationalstatus is available.

6.1.2 Categorisation ofoverqualification

Because of the superiority of the subjectiveapproach to the ‘objective’ DOT/GED system,as discussed in point 4.1.2 above, the strat-egy presented below for the measurement ofoverqualification is based on the former. The

61 Szydlik writes, ‘Admittedly, knowledge and skillsacquired abroad cannot be unequivocally evalu-ated’ (p.664); however, the continuation of hisstudy, which compares the job/training matchesof Germans and foreigners, is evidence of such anevaluation being undertaken in spite of this ac-knowledged obstacle; indeed, the evaluation is atthe very heart of this particular study.

62 For the rationale behind this exclusion, seeBüchel, 1998b.

Felix Büchel

506

strategy takes account of the objections thathave been raised about the subjectivity of thecollected information on job-requirement lev-els by validating reported information on oc-cupational status.

This generation of the mismatch variablesfrom three, rather than the usual two, initialindicators is what gives this categorisationsystem its innovative quality. It does, admit-tedly, create two additional categories: im-plausible combinations (of job-requirementlevel, formal qualification and occupationalstatus) as well as a ‘grey area’ (Hecker) or‘mixed category’ (Plicht et al., 1994), in whichthe three indicators referred to above cannotclearly signal a match or mismatch, eventhough there is nothing recognisably incon-sistent in the data reported by the respond-ent.63

One problem with the measurement strategyis therefore that the scope of this new mixedcategory scarcely lends itself to evaluation inthe framework of overqualification research.This disadvantage is more than offset, how-ever, by a sharper distinction between ad-equate employment and overqualification.64

The GSOEP does not ask directly whether arespondent’s job is commensurate with his orher training but merely asks about the re-quirements of the person’s job. The question‘What type of training is normally requiredfor the job you do?’ is accompanied by the fol-lowing graded response categories: ‘No spe-cial training required’, ‘Only a short induc-tion on the job’, ‘A lengthy period of coaching

at my place of work’, ‘Attendance at specialtheoretical or practical courses’, ‘A certificateof vocational training’ and ‘A university orcollege degree’. It is permissible to choosemore than one response. Pre-tests, however,have shown that multiple responses, whichare in any case quite rare, almost invariablyinvolve a combination of ‘A certificate of vo-cational training’ and ‘A lengthy period ofcoaching at my place of work’ or else the re-quirement category ‘Attendance at specialtheoretical or practical courses’ in conjunctionwith any of the other options. This means thatwe can record the most demanding reportedrequirement without sacrificing any very sig-nificant information.65

By means of a comparison with the acquiredformal vocational-training qualification,66 thedegree of congruence between job and educa-tion can be directly determined; the dispar-ity between the employee’s formal qualifica-tion and the job requirements – at least forpeople with the same category of formal quali-fication – can then be graded, i.e. assigned toone of a number of bands in a ‘skill spill’ scale.Since the pre-tests that were conducted aspart of the data-verification process high-

63 To define the mixed category, we must first con-duct pre-tests (which are not documented here),using group-specific occupational and income in-formation.

64 A further residual category, created by the pres-ence of a missing value in one of the three indica-tors required to generate the mismatch variables,is no different in methodological terms to anequivalent category based on a conventional two-variable approach but is made somewhat largerby the addition of the third variable. The informa-tion on occupational status is not regarded as sen-sitive data and is seldom posted as a missing valuein the GSOEP.

65 In conjunction with the income distribution foreach category and the structure of the job classifi-cations in each category (on the basis of the three-digit ISCO codes), summaries were produced inadvance for each of the requirement levels. Thesepre-tests revealed, for example, that the occupa-tional and income structure for the two require-ment levels ‘A certificate of vocational training’ and‘Attendance at special theoretical or practicaltraining courses are similar, in the same way asthe structures for ‘No special training required’ and‘Only a short induction on the job’ resemble eachother. For that reason, each of these response pairsis bracketed together.

66 Following similarly structured pre-tests, the lev-els of formal qualification are aggregated. The cer-tificate of vocational training, which is the gen-eral category for those workers whose qualifica-tions are non-academic, covers those who are quali-fied master craftsmen, as is common practice;graduates of universities and colleges of highertechnical education (Fachhochschulen) are alsobracketed together in the customary manner,which reflects the fact that no distinction is madebetween these two levels of qualification on thejob-requirement side.


507

lighted individual borderline cases straddlingtwo categories as well as a number of incon-sistencies, other particularly detailed infor-mation on the respondents’ occupational sta-tus from the GSOEP survey is included inorder to determine the mismatch variables.

The following categories are thus created:‘(definitely) overqualified’, ‘degree of mis-match not clearly determinable’ (a mixed cat-egory which is wisely excluded as a rule fromevaluations of job content), ‘(definite) congru-ence between job and education’, ‘implausi-ble combination of the three generating vari-ables’ and ‘at least one missing value amongthe three generating variables’ (an indeter-minable degree of mismatch is also classedas a missing value).67

The use of separate classification models forwestern and eastern Germany takes accountof the different educational certificates in theformer East and West Germany.68 The classi-fication model for western Germany with thecombinations of job requirement, formalqualification and occupational status is docu-mented in Table 0 of the annex; for the classi-fication models for eastern Germany, seeBüchel, 1998b, Tables 4.2 and 4.3, pp.189 ff.

Identifying underqualification

Occasionally, when analysing the job/qualifi-cation fit, besides examining the category ofadequate employment and that of overqualifi-cation, researchers also study a second misfitcategory, namely that of underqualification(see for example Daly et al. (forthcoming)).Such a situation occurs when a respondentdoes a job for which he or she does not pos-sess adequate formal qualifications. This mis-match phenomenon occurs in a few excep-tional ‘rags-to-riches’ careers, such as that of

a dishwasher who works his way up to be-come a millionaire, but it has to be discussedon the basis of completely different premisesto those that underlie the discussion ofoverqualification (cf. Alpin et al., 1998, pp.19-20). The theory of career mobility, for exam-ple, can explain overqualification but notoveremployment; Sicherman (1991, p.109)explicitly addresses this point.69 If research-ers wish to examine mismatches in general,they must accordingly analyse and discuss thecategories ‘overqualification’ and ‘underquali-fication’ in strict segregation. If, on the otherhand, the emphasis is on the phenomenon ofoverqualification, as it is here, underqualifi-cation and adequate qualification can belumped together fairly easily (for a similarapproach, see for example Groot, 1993a,p.302, and Groot, 1996). In the following para-graphs, therefore, underqualified workers areassigned to the ‘adequate employment’ cat-egory; their common quality that interests usin the present context is the fact that theyare identified as not being overqualified.

The degree of overqualification

The aforementioned grading scale for data(recoded into categories) on formal vocationalqualification and job requirement can also beused to interpret and grade the newly createdinformation on the discrepancy between thesejob and training levels. However, when thecategorisation is subjected to further differ-entiation in this way, since the original vari-ables were already highly aggregated, theresultant lack of sharp divisions between thecategories demands a certain amount of cau-tion from anyone who develops such a classi-fication model or who subsequently interpretsthe findings it produces.

67 For a treatment of the relatively small percent-ages that fall into this category and into those of‘implausible combinations’ and ‘missing values’,see Büchel and Weißhuhn, 1997a, Tables A2-W/O-84, -91 and –93, as well as Büchel and Weißhuhn,1998, Table A2-W/O-95.

68 For the rationale behind this division, seeBüchel, 1998b.

69 Moreover, in European countries such as Ger-many, where certified vocational qualifications aremajor determinants of career prospects – in con-trast to the United States, for example – under-qualification is of minimal empirical significance.The 7% or so of German employees (1984 figure)whose educational certificates would not normallyqualify them to pursue their current occupationhave an average shortfall of only 0.7 years of school-ing; the average qualification surplus for overquali-fied workers in 1984, on the other hand, exceededtwo years (see Daly et al., 1997, Table 1, p.27).

Felix Büchel

508

While the division into appropriate employ-ment and overqualification on the basis of thedescribed categorisation system has certainlypassed the validity test (cf. point 6.1.2 above),this cannot be said unreservedly of the de-gree of overqualification. Although there is nodoubting the validity of the graded structure,the volume of the various skill spills cannotbe determined with sufficient accuracy. Forthat reason the graded structure has onlybeen used to obtain information on the vol-ume of skill spillage in particular descriptiveevaluations where incidence figures havebeen high (see for example Büchel, 1998b,Tables 5.1 to 5.3, pp.192 ff.).

For more ambitious research designs, such asthe one illustrated in the present section, ittherefore seems advisable to accept some lossof information for the sake of a more validcategorisation system; this, indeed, is donein most empirical studies on overqualification.Thus, if the sample is classified into valid cat-egories, the result will be a dummy variable– overqualified: yes/no – for all workers witha formal vocational qualification or a univer-sity degree acquired in Germany.

Validity checking

The overqualification variable obtained on thebasis of the categorisation principle describedin the preceding paragraphs must be exter-nally and internally validated before beingused empirically.

The external validation is based on the com-parability of the volume of underqualificationrevealed by one’s own findings with the fig-ures produced by other studies with differentunderlying research strategies.

The most comprehensive empirical study ofthe incidence of overqualification in Germanywhich is not based on the categorisation strat-egy described above is that of Plicht et al.(1994), although it only examines universitygraduates. The study involves a mixed cat-egory, which results in wide margins of error;within this framework it identifies graduateoverqualification ranging from 8% to 17%(western Germany, 1991). The author of thepresent study, eliminating the mixed category,

arrived at a value of about 14% (see Büchel,1998b, Table 5.2, p.193), which is of the sameorder as Plicht, Schober and Schreyer, eventhough they not only use a different databasebut also pursue an entirely different ap-proach, based on occupational categories.

In studies on overqualification among univer-sity graduates, which is the main focus of at-tention in the German discussion, percent-ages of under 10% to 15% (Tessaring, 1994a,p.49, Buttler and Tessaring, 1993, p.470, andSzydlik, 1996b, p.304) are identified, althoughno details are given on any part of the opera-tions used to arrive at these percentages. Thefigure of 25% advanced by the Donors’ Asso-ciation for the Promotion of Science and theHumanities in Germany (Stifterverband fürdie Deutsche Wissenschaft, 1993, p.4) issurely overestimated and must be seen in thelight of the general pessimistic tone of thatpublication. So if we exclude this last outlier,the external validation seems to have beengenerally successful.

For the purpose of internal validation, thevalidity of the construct used to distinguishbetween the characteristics of jobs done byappropriately qualified workers and thosewhose incumbents are overqualified is tested.If the categories really are sharply defined,considerable differences should emerge interms of quality.

Irrespective of its theoretical derivation, theexpectation that overqualified workers wouldbe considerably worse off than adequatelyemployed workers with similar formal quali-fications is observably fulfilled in a highly re-fined examination of some 50 items in virtu-ally all dimensions (Table 1 in the annex). Thepattern of these findings leaves no room fordoubt about the validity of the categorisationmodel.

6.1.3 The evaluation process

The empirical evaluations are based on theGerman Socio-Economic Panel (GSOEP – seepoint 6.1.1 above). West and East Germanyare examined separately. The cross-sectionevaluations are based on the year 1995, whilethe longitudinal analyses use the period from


509

1984 to 1995. Specific details of the analyti-cal methodology are contained in the variousparagraphs on the interpretation of findings.The findings that are presented are largelyderived from the multivariate analyses. Forfurther descriptive information on the vari-ous stages in the evaluation process, thereader is referred to Büchel, 1998b. Over-qualification is identified on the basis of thestrategy described here in subsection 6.1.2.

Subsection 6.2 examines four subjective indi-cators that produce inferior values for joblesspeople compared with employees in the fieldof unemployment research, refining this strat-egy by applying a division into three catego-ries – ‘unemployment/hidden reserves’,70

‘overqualification’ and ‘adequate employ-ment’. Subsections 6.3 and 6.4 go on to ana-lyse transition patterns into and out ofoverqualification. Finally, section 6.5 exam-ines the dynamics of income loss resultingfrom overqualification; the second part of thisevaluation focuses on a comparison betweenlosses resulting from periods of unemploy-ment and periods of overqualification.

6.2 Subjective indicators: differencesbetween adequate employment,overqualification and unemployment

6.2.1 Problem-solving capacity

In order to measure possible deprivation inpeople’s everyday lives, the questionnaireasks respondents to what extent they agreewith a number of statements, one of which is‘Things have become so complicated that I canbarely cope’. Although this question has a rec-ognisable dynamic component, which relatesto external (social) development, the responsecan also apply to transitions from a sociallysecure status, such as training or adequateemployment, to overqualification or unem-ployment and therefore seems to be a very

suitable means of shedding light on the issueunder examination here. The responses ‘fullyagree’ and ‘tend to agree’ with the aforemen-tioned statement are interpreted as indica-tive of limited problem-solving capacity. Aprobit model is used to identify the determi-nants of this limited capacity, special atten-tion being focused on the three relevant cat-egories of employment status.71

The findings set out in Table 2 of the annexshow that the overqualified respondents’ as-sessment of their own problem-solving capac-ity, controlled for important socioeconomicvariables, including the level of formal edu-cation, actually places their expected ‘middle’position far closer to that of jobseekers thanto adequately employed workers. Althoughtheir problem-solving capacity is higher, onaverage, than that of jobseekers, the differ-ence is not significant. By contrast, there aresignificant differentials between matched andmismatched employees. This finding is ob-servable for both West and East Germany.

6.2.2 Morale

A person’s awareness of a shortage or absenceof demand in the labour market for his or heracquired vocational skills and qualificationswill probably have an adverse effect on thatperson’s morale, since the vast majority ofpeople, or at least of those who wish to be eco-nomically active, place employment high ontheir list of priorities. For the purposes of thestudy, the sample is divided into those whosemorale is high and others. In accordance witha strategy commonly used by the GSOEP toevaluate morale, based on an empirical dis-tribution of responses on an eleven-point scale(0-10), people who rate themselves at points8, 9 and 10 on the scale are defined as havinghigh morale.72

70 Hidden reserves of labour are those non-work-ing members of the active population people whoare not registered as unemployed but who reportimmediate readiness to take up employment. Forthe sake of linguistic simplicity, the ‘unemployedor hidden reserves’ category is abbreviated to ‘un-employed’ in the remainder of this section.

71 The same type of econometric model is adoptedin points 6.2.2 to 6.2.4 below.

72 American researchers occasionally display lesssensitivity in their choice of evaluation model.Hersch (1991), for example, evaluates her satis-faction scale, which is graded in the same way asthe GSOEP model, on the basis of a regressionequation.

Felix Büchel

510

The following may be deduced from the find-ings set out in Table 3 of the annex: both inWest and East Germany, overqualified work-ers report not only significantly higher mo-rale than jobseekers but also significantlylower morale than those workers whose jobsmatch their qualification levels. Whereas theoverqualified workers are roughly equidistantfrom the other two categories on the satisfac-tion scale for East Germany, the morale ofoverqualified workers in West Germany is farcloser to that of adequately employed work-ers than to that of jobseekers.

6.2.3 People’s concern about their owneconomic prospects

In order to measure the respondents’ assess-ment of their own prospects for the future,the questionnaire asks whether they are wor-ried about a number of things, one of whichis their own economic situation. The responses‘very worried’ and ‘rather worried’ are inter-preted as pessimistic assessments of the re-spondents’ own economic prospects.

The findings are set out in Table 4 of the an-nex. Both in West and East Germany,overqualified workers are significantly moreoptimistic about their expected economic de-velopment than jobseekers but significantlymore pessimistic than workers with well-matched jobs. One interesting point is thatthe same pattern emerges as was observedfor morale, with equal spacing between themiddle category and each of the other two forEast Germany, whereas overqualified work-ers in West Germany are distinctly closer toadequately employed workers than to job-seekers in their assessment of their own eco-nomic prospects.

6.2.4 Political involvement

An important observation in the domain ofpolitical science is that unemployed personsare liable to turn their backs on the world ofpolitics because they feel betrayed (the ‘po-litical apathy’ hypothesis; see Büchel andFalter, 1994). This is liable to lead to in im-paired functioning of the democratic system.Besides voting in elections, allegiance to apolitical party is regarded as an effective in-

dicator of the extent of a person’s political in-volvement. The purpose of these paragraphsis to extend the conventional examination ofthe connection between employment statusand party allegiance to the category of unem-ployed workers.

The findings documented in Table 5 of theannex show that, controlling for major socio-economic characteristics, the identification ofjobseekers with the party system does notdiffer significantly from that of overqualifiedworkers. The latter, however, demonstratesignificantly less party allegiance than work-ers with good job/qualification matches.

All in all, the findings of section 6.2 confirmthe hypothesis that the redundancy of an in-dividual’s marketable skills does not onlyhave an adverse effect on his or her social well-being in the case of total redundancy (i.e.unemployment). Even partial redundancy ofa person’s skills, as happens in the event ofoverqualification, results in a diminished –and sometimes drastically diminished – qual-ity of life compared with that enjoyed by em-ployees whose jobs are commensurate withtheir qualifications. While overqualified work-ers are significantly better placed thanjobseekers and significantly worse off thantheir adequately employed counterparts interms of morale and perception of their owneconomic prospects, once the relevant datahas been controlled for important socioeco-nomic characteristics, when it comes to theirability to cope with everyday life (at least forthe time being) and their allegiance to politi-cal parties, however, overqualified workersdemonstrate attitudes that are not far re-moved from those displayed by jobseekers,whereas people whose jobs fit their qualifica-tions tend to be significantly closer to the ‘so-cial norm’ in these respects.

6.3 Transition to inadequateemployment

6.3.1 Preliminary remarks on the dy-namics of individual overqualification

Following the presentation of the static per-spective in the previous subsection (for awider range of analyses see Büchel, 1998b),


511

the following paragraphs deal with the dy-namics of changes of employment status atan individual level. Although in the mediumterm the percentage of overqualified employ-ees in the German labour force has remainedfairly stable (see Büchel and Weißhuhn, 1997aand 1998), we cannot rule out the possibilitythat, as with the phenomenon of unemploy-ment, this overall picture of stability concealsa great deal of movement in and out ofoverqualification in the course of individualcareers (cf. Groot and Maassen van den Brink,1995, p.14).

Accordingly, the approaches adopted here arebased on the longitudinal analyses that char-acterise modern unemployment research.This means that interest focuses on the phe-nomena of movements into and out ofoverqualification, which have seldom beendealt with in research literature, as well ason income effects of overqualification. Themain points of interest are transfers in bothdirections between overqualification and theother main categories of employment status,namely work commensurate with qualifica-tions, unemployment, other economic inactiv-ity and full-time training. The main transi-tion patterns are examined below.73 Whenchoosing suitable evaluation procedures, wehave to remember that the information on job/training congruence is only available for onepoint in the year, in other words the time ofthe survey. Since some of the observation pe-riods are very short (1991 to 1995 for the areaof the former East Germany, as against 1984to 1995 for western Germany), the use of haz-ard-rate models, which is actually the bestway to proceed when dealing with this sort ofquestion, would be ill-advised, even for dis-crete modelling. The use of the panel approachto process the relevant data is also out of thequestion at almost every stage of the evalua-tion, since the observed changes of employ-ment status will normally occur only once, ifat all, for each person during the examina-tion period. Accordingly, models that controlfor unobserved heterogeneity would excludemost transitions from the examination, since

observations with only one occurrence do notserve any explanatory purpose in this type ofmodel. For that reason, the research designsof the various stages in the evaluation proc-ess must be conceived in such a way that theanalysis can be performed with conventionalprobit or regression models.

Our first step is to examine the processwhereby individuals become overqualified.This transition is identified when a respond-ent reports overqualification in a question-naire after having reported a different em-ployment status the previous year. Thesesituations from which individuals enter intooverqualification are the focal point of thepresent subsection.

6.3.2 Descriptive findings

Table 6 in the annex shows the types of em-ployment status previously held by overquali-fied workers, broken down into percentagecolumns by regional labour market, qualifi-cation level and sex.

An examination of the totals column showsthat the most common previous status ofoverqualified workers in both East and WestGermany is that of adequate employment(just over 40% in each case). Outdated exper-tise and declining productive capacity may bemooted as reasons for this.74 This transitionpattern is far more characteristic of men thanwomen, reaching a peak for highly qualifiedmale workers in the former East Germany;this reflects a systematic downgrading proc-ess as part of the hierarchical ‘cleansing’ op-eration in the period following reunification.Nevertheless, this seems to refute the oft-re-peated hypothesis that overqualification ischiefly a problem for married women who in-terrupt their careers for family reasons, withthe result that their skills become rustythrough lack of practice, leaving them no op-

73 For more findings see Büchel, 1998b.

74 See the analyses, broken down by age group, inBüchel and Weißhuhn, 1997a and 1998, whichdemonstrate a considerably higher percentage ofoverqualified employees in the oldest age bracket,especially among non-graduates in western Ger-many.

Felix Büchel

512

tion but to re-enter working life at a level forwhich they are overqualified; while this ex-plains some overqualification, it is clearly notthe only explanation.

In western Germany, almost a quarter of thosewho take up underskilled work are unem-ployed jobseekers. In the East, this percent-age is considerably higher, with some 40% ofoverqualified recruits having been unem-ployed the year before. This demonstrates thenature of overqualification as an alternativeto unemployed status; even though privateplacement services for unemployed people aregradually gaining in importance, this sizeablemovement of labour from unemployment intooverqualification is due not least to the re-striction by the national employment author-ity of the jobseeker’s right to refuse job of-fers. The findings also testify to an abundanceof flexibility on the part of jobseekers. Theyalso reveal striking differences between re-gions and between levels of formal qualifica-tion; for example, underskilled jobs are a veryimportant source of employment for EastGerman workers with mid-range qualifica-tions, whereas for West German graduates,especially males, overqualification has far lessempirical significance.

‘Spontaneous’ transitions from economic in-activity to overqualification by people who arenot actively seeking work (i.e. who are notregistered as unemployed) are only made toany great extent by West German women withmid-range qualifications. West Germanwomen graduates lag far behind in secondplace. Among West German men and EastGermans in general, this form of transitionhas only marginal significance.

Approximately one in eight people take up anunderskilled job on completion of a trainingcourse. This is the case in both western andeastern Germany. This type of transition hasbecome disturbingly common among WestGerman graduates, both male and female.More than one in four West German gradu-ates working overqualified took up hisunderskilled job immediately after obtainingtheir degrees; this situation is somewhat lesscommon in eastern Germany and applies pri-marily to highly qualified women.

6.3.3 Multivariate findings

The reference category for our multivariateanalysis was the length of time spent in oneof the other types of employment status be-sides overqualification. The multivariateanalysis of the descriptively acquired findings(Table 7 in the annex) produces a surprisingpicture in terms of the previous employmentstatus of overqualified workers. Our first task,however, is to discuss the findings in relationto the control variables.

Clearly different behaviour patterns emergefor transitions to overqualification in easternand western Germany.

In western Germany, first of all, more womenthan men take jobs for which they areoverqualified. Reasons for this may be lessemployment-centred training courses, lowerpriority attached to careers, periods of eco-nomic inactivity for family reasons with theaccompanying depreciation of human capital,etc. (see below). Older employees are seldomobserved to make such a switch. It may beexpected that, on losing jobs commensuratewith their qualifications, for example, olderemployees may be more inclined to take ad-vantage of early-retirement packages than to‘struggle through’ to retirement in anunderskilled job.

Transitions to overqualification are more fre-quently observed among individuals with poorschool records than among those with betterschool qualifications. Training in a technicalcollege (Fachschule), which has traditionallybeen employment-based, affords protectionagainst overqualification, as does civil-serv-ice training, which is normally followed byappointment to an appropriate post in one ofthe public authorities. Individuals who havesuccessfully completed an apprenticeship arealso less liable to become overqualified thanthose who have undergone other forms of vo-cational training.

A surprising result emerges for graduates ofhigher technical colleges (Fachhochschulen),who are observed to be more prone tooverqualification than holders of apprentice-ship certificates. Respondents who attach low


513

priority to a successful career are also over-represented among the newly overqualified;the undemanding nature of their jobs corre-sponds to their own level of demand for pro-fessional success.

As the regional unemployment rate rises, theprobability of a transition to overqualificationis reduced. There are two conceivable waysof explaining this: either unskilled and semi-skilled jobs are disproportionately rare in atight job market or else individuals who can-not obtain an appropriate job are more in-clined simply to register as unemployed, thesocial stigma of unemployment being dimin-ished as the local jobless rate increases.

In the territory of the former East Germany,the control variables exert an influence inseveral different ways. Besides older employ-ees, those in the youngest age bracket alsomove into overqualification less frequently,which means that transitions into overquali-fication are typically made by workers in themiddle age groups. While there are no gen-der-specific differences of the sort that existin western Germany, people who only havethe certificate awarded on the successful com-pletion of eight years’ schooling are also morelikely to be overqualified. Another conspicu-ous feature is the over-representation ofhighly qualified employees, who were the vic-tims of large-scale systematic downgrading.The reason given for this measure by employ-ers is the discrepancy between the high levelof employees’ formal qualifications, which of-ten depended on a certain degree of identifi-cation with the former Communist system,and the actual market value of these qualifi-cations in a competitive economy. In easternGermany too, transitions to overqualificationtend to occur more frequently in areas withrelatively low rates of unemployment. Moreo-ver, such transitions are more common in ru-ral than in urban areas, in contrast to theolder federal states.

As for the previous employment status ofoverqualified workers, once the quite hetero-geneous descriptive findings have been con-trolled for important socioeconomic character-istics, surprisingly similar structures emergefor eastern and western Germany. Compared

with a transition from appropriate employ-ment – the reference category – a switch fromunemployment or the hidden labour reserveis a far more frequent occurrence. Direct tran-sitions from training are also significantlymore common than those from appropriateemployment, albeit by a somewhat narrowermargin. In the west of Germany, moreover,even transitions from other forms of economicinactivity happen more often than transitionsfrom a matched to a mismatched job; in theeastern part of the country, however, otherforms of economic inactivity play an insignifi-cant role as a source of overqualified labour.

This shows that the descriptively identified‘typical’ switch from a well-matched job to amismatch correlates highly with major socio-economic characteristics. Controlling for these,we see that this form of transition is actuallyrather untypical. Hence, overqualification inGermany does not primarily take the form ofworker dequalification, for example in thewake of rapid technological change, with theaccompanying devaluation of knowledge andskills; on the contrary, underskilled jobs tendto be accepted by those who are economicallyinactive in whatever form. The motives, whichare surely many and varied, for taking suchjobs are still uncertain at this stage of theanalysis, but more precise information on thesewill emerge on further examination.

6.4 Selected transitions from over-qualification

Our second step involved an examination ofthe process whereby individuals cease to beoverqualified. Here too, the various forms ofemployment status that were discussed inconnection with transitions to overqualifi-cation were the main focus of attention. Par-ticular importance attached to the questionwhether overqualified workers manage tostep up into jobs that are commensurate withtheir training or whether they are more likelyto slip down into unemployment.

Descriptive findings

The findings presented in Table8 of the an-nex reveal considerable differences betweeneastern and western Germany.

Felix Büchel

514

The most commonly used exit door fromoverqualification in western Germany leadsto appropriate employment. A conspicuouslyhigh number of male graduates make thattransition. While this can be used as an indi-cator of the capacity of the career mobilitytheory to explain overqualification, it is notevidence in itself, since we are only lookinghere at those who move away from under-skilled jobs, not at those who remain in them.A very similar pattern emerges, albeit at alower level, in eastern Germany, but this timethe main exit route leads to involuntary eco-nomic inactivity, i.e. unemployment or a placein the hidden reserve of labour. This type oftransition mainly affects people with mid-range qualifications as well as women withhigher levels of qualification. In the west ofGermany this downward transition is mainlyobservable among men with vocational diplo-mas; it hardly ever affects male graduates andseldom happens to female graduates.

There is also a wide East/West disparity inthe case of transitions into voluntary economicinactivity. In western Germany, this transi-tion is typically made by women who give upoverqualified work. Irrespective of their levelof qualification, some 40% of women who giveup underskilled jobs move into voluntary eco-nomic inactivity. In eastern Germany, how-ever, this type of transition is not very sig-nificant. Percentages vary little among thecompared groups, reaching a peak of 10%.

A switch to full-time training is not a charac-teristic move for those who give up inadequateemployment. Neither in the East nor in theWest of Germany does this figure reach 10%.Whereas in eastern Germany women aremore likely to make such a move, whereas inthe West German labour market this transi-tion is chiefly made by men, with holders ofacademic degrees more frequently being giventhe opportunity for full-time training (pre-sumably in the form of a second degree courseor professional training) than holders of vo-cational diplomas.

The following paragraphs will take a closerlook at transitions from overqualification tothe two types of status that are of most inter-est to us in the context of this study, namely

unemployment and adequate employment.For more detailed analyses, see Büchel,1998b.

6.4.1 Transition to unemployment

At this juncture a two-step evaluation wasundertaken with a view to answering the fol-lowing questions:

a) Do people in underskilled jobs run agreater risk of unemployment than thosein appropriate jobs?

b) Does the acceptance of an underskilled jobprotect people who have lost an appropri-ate job from subsequent unemployment?Or, to put it another way, does overqualifi-cation offer such people a genuine alter-native to unemployment in the longerterm?

Transition to unemployment frominadequate or adequate employment

The aim here was to examine whether ad-equate employment affords better protectionthan overqualification against the risk of un-employment. In the model that was used, thealternative to a transition to unemploymentwas continued overqualification.

The findings set out in Table 9 of the annexreveal considerable differences between west-ern and eastern Germany in terms of the riskof losing a job and becoming unemployed.They do concur, however, on one key point,namely that this risk is significantly higherfor employees in underskilled jobs than forthose whose jobs match their qualifications.The respective risk levels for these two groupsdiverge more widely in the East than in theWest of Germany.

Transition from adequate employment tounemployment – directly or viaoverqualification

We wished to test whether the acceptance ofan underskilled post by a person who had losta job commensurate with his or her skills andqualifications was likely to protect that per-son against subsequent unemployment. To


515

paraphrase the question, we want to knowwhether overqualification is a genuine alter-native to unemployment or whether it is li-able only to postpone a descent into unem-ployment. Those people who remained inadequate employment over the observationperiod served as the reference category.

What the findings in Table 10 of the annexdemonstrate is that people who take anunderskilled job on losing an adequate jobconsiderably reduce the risk of subsequentunemployment, particularly in western Ger-many. This may be deemed to indicate thatacceptance of an underskilled job is a delib-erate choice on the part of many people as analternative to unemployment.

6.4.2 Transition to adequate employ-ment

At this point a two-step evaluation was un-dertaken with a view to answering the fol-lowing questions:

a) Are individuals more likely to obtain a jobcommensurate with their qualifications ifthey are overqualified or if they are unem-ployed?

b) Does an interim phase of overqualificationmake it easier for jobseekers to find a postthat matches their qualifications?

Transition to adequate employment fromoverqualification or unemployment

The purpose of this stage in the evaluationprocess was to establish whether the fact thatoverqualified people already have one foot inthe labour market gives them an advantageover their unemployed counterparts in thequest for adequate employment, assumingthey are in competition for the same jobs. Peo-ple who are economically inactive but not reg-istered as unemployed or in the hidden reserve,or who have just completed a course of train-ing were excluded from this evaluation. Forunemployed jobseekers and overqualifiedworkers together, we examined the probabil-ity of transition to adequate employment. Thereference category comprised those who re-tained one status – unemployment or over-qualification – throughout the observation pe-

riod. Our primary interest was to discoverwhether and how a person’s original statusinfluenced his or her chances of making thetransition to adequate employment. The find-ings are presented in Table 11 of the annex.

Overqualified workers find adequate jobs lessfrequently than unemployed jobseekers. Thesame pattern applies in both parts of Ger-many. It does not seem easy to interpret thesefindings. It is conceivable that people inunderskilled jobs are under less pressure tofind appropriate jobs than those with no jobat all, since the job they have, inadequatethough it may be, probably provides themwith an income above that of unemployedjobseekers. Moreover, as has been shown, thecategory of overqualified workers includessome people who attach low priority to a suc-cessful career; it is conceivable that such peo-ple may not be looking for a job that is morecommensurate with their qualifications or atleast that their efforts to find a better job maybe less than wholehearted.

To put it simplistically,75 it may be said ofunemployed jobseekers that their status tendsto ‘testify’ to a disinclination to accept any oldjob – including, of course, underskilled jobs;it is therefore safe to assume a high degree ofprobability that their job search will be con-fined to educationally adequate employment,at least in the initial stages. What certainlyemerges is that, when employers have jobvacancies, they do not seem to let their choiceof candidate be governed to any great extentby the stigmatisation effect of unemploymentor by the positive signal of willingness to work‘at all costs’ that is transmitted by applicantswho are overqualified for their current jobs.

Transition from unemployment to adequateemployment – directly or via overqualification

The next step was to undertake a more spe-cific examination of the findings described inthe previous paragraphs. To this end, job-seekers were studied with a view to ascertain-

75 ‘Simplistically’ in the sense that we are assum-ing the availability of unskilled jobs at minimalwage rates.

Felix Büchel

516

ing whether they managed to find education-ally adequate employment within a particu-lar period – five years for western Germanyand three years for the territory of the formerEast Germany. This means that transitionpatterns could be compared between the twoparts of Germany but not the frequency oftransitions. The question at the forefront ofthis examination was whether a transitionalphase of overqualification is liable to serveas a ‘bridgehead’, in other words whether ithelps jobseekers to gain a foothold in the la-bour market from where it is easier for themto move upmarket into the sort of job for whichthey are qualified (for a more general treat-ment of this question, see Büchel, 1994a).

The findings set out in Table 12 of the annexsuggest that, if adequate employment is as-sumed to be the ultimate aim of a job search,a temporary ‘makeshift’ transition to over-qualification tends, in both the East and theWest of Germany, to be a ‘dead end’. The di-mensions of the estimated parameters indi-cate that a high volume of state dependenceensues from this situation.

6.5 Income effects of overqualification

There is already sufficient statistical evidencethat overqualified workers earn lower in-comes than workers with equivalent qualifi-cations whose jobs match those qualifications(see section4 above), and this point need notbe analysed any further here. What this sub-section examines is the extent to which in-come growth differs between overqualifiedand appropriately employed workers.

6.5.1 Income growth among people whoare initially overqualified

The first step was to compare the respond-ents’ first reported income in the observationperiod (broken down by degree of job/train-ing match) with their last reported income;in the case of the latter figure, the match/mis-match question was no longer relevant, be-cause the analysis took account of any upwardmobility from underskilled jobs.

In the evaluation, the initial amount of in-come was taken into consideration so that we

could control for the fact that the lowest earn-ers automatically stand a better chance ofimproving their income levels. The ratio of thelast to the first reported income was selectedas the dependent variable; these values wereplotted on a logarithmic scale to adapt themto the model. Given the nature of the depend-ent variables, extraneous control for thelength of the observation period was also re-quired. In the vast majority of cases, the du-ration of this period equalled the number ofyears spent in employment. To control foruntypical career histories, intervening peri-ods of unemployment and other forms of eco-nomic inactivity were taken into account asadditional covariates,76 as were any yearsmissing from the panel records.

The findings set forth in Table 13 of the an-nex present a very similar picture for easternand western Germany in terms of the effectof adequate employment at the start of theobservation period. Controlling for the initialincome level, we see that the income curvefor overqualified workers rises less steeplythan for workers whose jobs match their train-ing levels. So the former not only have to en-dure an income shortfall at any given time inrespect of their unused skills and qualifica-tions (see the literature cited in section4 aboveor, for Germany, Büchel, 1998b, Table 5.30,p.238, for example); in addition, the gap be-tween their earning power and that of theiradequately employed colleagues widens withlength of service.

6.5.2 Erosion of human capital throughoverqualification

The second step involved examining whethera period of overqualification has the same sortof effect in terms of depreciation of humancapital as a phase of voluntary or involuntaryeconomic inactivity. This hypothesised affin-ity is based on the fact that 100% of a per-son’s acquired human capital lies dormantduring a phase of economic inactivity, whilex% of it lies dormant during a phase of

76 In this context, ‘number of years’ means thenumber of annual surveys in which a particularemployment status was reported.


517

overqualification, x being situated somewherebetween 0 and 100. Accordingly, the rate ofskill depreciation resulting from overqualifi-cation must lie somewhere between the de-preciation rate for an economically inactiveperson and the rate for a person in adequateemployment.

Clearly, this hypothesis can only be tested iffigures for initial and final income from ad-equate employment are available and if aphase of overqualification is identifiable atsome time between the beginning and end ofthe observation period. In order to permit adirect comparison with the standard referencevalue, i.e. a period of unemployment, peoplewith a history of adequate employment fol-lowed by a period of unemployment or a pe-riod in the hidden labour reserve then anotherperiod of adequate employment were includedin the analysis. As in the previous step, thereported data were controlled for interveningyears of voluntary economic inactivity and forany years that were missing from the panelrecords.

The findings presented in Table 14 of the an-nex confirm the aforementioned hypothesis,at least for western Germany.77 In both partsof Germany, periods of overqualificationsandwiched between periods of adequate em-ployment adversely affect income growth. Inthe West, as expected, the depreciation rateis lower than that for voluntary inactivity(100% absence of training in occupationalskills), which in turn is lower than the de-preciation rate for unemployed jobseekers(100% non-use of skills plus additional down-ward pressure on wage demands). This ac-cords with the expectations of the human-capital theory and may also be regarded astending to confirm the typological affinitybetween overqualification and unemploy-ment that was postulated in the introduc-tion to the present study.

7. Conclusions

7.1 Methodological consequences foremployment researchers

Conventional employment research, with itsfixation on unemployment, underestimatesthe volume of skills and qualifications whicheducation systems produce but for whichthere is no demand in the labour market. Byincluding the hidden reserve of labour inanalyses of this type,78 employment research-ers have already taken one major step to-wards more effective assessment of the effi-ciency of education systems and of thecoordination problem between education sys-tems and labour markets. The purpose of thepresent study has been to highlight the ap-parent indispensability of another step,namely consideration of overqualification asan additional category in its own right.

While the category of the hidden reserve hasnow found its way into official and semi-offi-cial statistics and has secured a permanentplace alongside registered unemployment (seefor example OECD, 1995b, pp.45 ff.), eitherthe overqualified percentage of the labourforce is tacitly omitted (as in the official Ger-man employment statistics) or its omission isascribed to unresolved measurement prob-lems: ‘Invisible underemployment [i.e. thepercentage of the labour force who areoverqualified for their jobs] … by its very na-ture is difficult to measure. For this reason,it is not discussed’ (OECD, 1995b, p.45).

A first attempt to close this unsatisfactoryinformation gap was made by the GermanMinistry of Education, Science, Research andTechnology, which began to commission bien-nial standardised reports on the extent andstructure of overqualification in Germany (seeBüchel and Weißhuhn, 1997a and 1998).

77 The unexpected findings for eastern Germanyhave to be interpreted in the light of the particu-lar situation that still obtains in the labour mar-ket there.

78 In this context, however, employment research-ers usually disregard the phenomenon of ‘visibleunderemployment’, i.e. the fact that numerousemployees would like to work longer hours (seeOECD, 1995a, p.45). This is another situation inwhich people are prevented from capitalising fullyon their skills and qualifications.

Felix Büchel

518

The measurement strategy used in these re-ports, however, relates closely to the specificcharacteristics of the German education sys-tem and cannot answer Rumberger’s call forthe development of an internationally com-parable indicator (Rumberger, 1994, p.281).Given the wide diversity of national educa-tion systems, more could surely be achievedwith measurement strategies (less precisethough they would necessarily be) which werebased on training duration and could there-fore be integrated into an extended human-capital approach of the type described inDuncan and Hoffmann (1981).

The extensive omission of the phenomenonof overqualification from European employ-ment research creates problems, some ofwhich are critical, with regard to the validityof the applied examination method. This isespecially conspicuous in research into thedemand for skills and qualifications.

The problem lies in the almost universal prac-tice of implicitly equating the level of skillrequired for a job with the formal qualifica-tion of the incumbent. A typical statementappears, for example, in iwd (1997): ‘Low-skilled workers are the great losers of struc-tural change. Many unskilled and semi-skilledjobs in industry have already gone’ (p.6). Thisassimilation would only be valid if overqualifi-cation did not exist in the labour market. Aslong as it does exist – as it does in every in-dustrialised nation of the Western world – andas long as the incidence of overqualificationexceeds that of underqualification, there willbe more unskilled jobs than unskilled work-ers. Disregarding this fact can lead to vari-ous miscalculations, for instance to overesti-mation of the dynamics of structural changeand the accompanying loss of unskilled jobs.Likewise, the job prospects for unskilled andsemi-skilled workers will be systematicallypresented in an excessively gloomy light. Itmust be remembered that, alongside un-skilled and semi-skilled workers, there arealso overqualified workers in semi-skilled andunskilled jobs; the total number of all theseworkers represents the actual number of un-skilled and semi-skilled jobs in the labourmarket (and the economic demand for suchjobs). At the same time, the demand for higher

qualifications is systematically overesti-mated. Certainly the demand for overquali-fied workers proves that employers expect anoverqualified worker to outperform an ad-equately qualified worker in an equivalentjob. The fulfilment of this expectation can bedemonstrated empirically (Büchel, 1999a).Nevertheless, it remains undisputed that,where two workers have the same qualifica-tions, the one whose job matches those quali-fications will normally be more productivethan the one who is formally overqualified forhis job.

Although it may be the case that technologi-cal progress and the accompanying structuralchange are raising the average skill level re-quired for jobs in the European market, if thelevel of demand is deduced on the basis ofemployees’ formal qualifications, the educa-tion boom and the subsequent rise in the av-erage educational attainment level of the la-bour force will inevitably be the source ofmisinterpretations unless due account istaken of overqualification. Many differentsources of error affecting the accuracy of thepredicted demand for labour were quickly rec-ognised and discussed (see for exampleTessaring, 1982, or, for a more general treat-ment, Mertens, 1982, pp.145ff. and pp.563 ff.).However, the serious prediction problemsarising from the implicit assimilation of for-mal qualification and job requirement, prob-lems with consequences for educational plan-ning that must not be underestimated,especially in the domain of higher education,are still only being explicitly discussed in iso-lated studies (see Adamy and Bosch, 1990,p.118, von Henninges, 1996, p.78, andWeißhuhn 1996, p.86).

Conventional forecasts of educational require-ments, i.e. predictions based on the ex postdevelopment of the labour force’s formal quali-fication structure and extrapolating futureneeds on that basis (see Tessaring, 1994b,Weißhuhn, 1996, and Weißhuhn et al, 1994)plainly cannot do enough to satisfy the need ofprospective students and trainees for informa-tion on their future employment prospects. Onealternative is the production of studies on thefate of graduates in each institution, brokendown by disciplines. At least in the academic


519

field, the authors of these studies are gradu-ally moving away from the practice of assess-ing educational success on the sole basis of the‘employed’/’unemployed’ dichotomy and aregiving increasing consideration to the qualityof the jobs that employees do.

The body of literature on this subject, however,is still far from providing reliable and compre-hensive information for prospective studentsand trainees, even though researchers did nottake long to start thinking about strategies forthe provision of such information (see Cha-berny and Schober, 1982). There is no compre-hensive reporting system on all possible train-ing courses at every educational institution inEurope, nor are there any studies on the fateof graduates in the longer term or standard-ised measurement systems, without which theopportunities and risks of different trainingcourses cannot be compared, while there is alsoa fairly general lack of academic quality amongthe few studies that are available at thepresent time (cf. point 5.2.2 above).

Moreover, even with the present researchdesign, as in the case of conventional educa-tional needs forecasts, it is only possible todraw reliable conclusions for the future fromthe past if the basic conditions remain con-stant in the medium term. The prospects offinding a job with a particular qualificationare certainly a good example of the sort ofinformation that cannot normally be providedbecause of changes in these basic conditions.Prospective students and trainees would gen-erally be well advised to choose their courseson the basis of their intrinsic interests thanto be guided by predicted labour requirementsor the findings of studies on the employmentstatus of previous graduates. Be that as itmay, it is still a fact that demand predictionswhich do not address the problem of over-qualification are plagued with an even greaterdegree of uncertainty than those which areadjusted for overqualification.

On the basis of a few classic questions fromthe field of unemployment research, thepresent study has shown that the phenom-enon of overqualification, because of its typo-logical affinity with unemployment, can beanalysed effectively with the instruments of

unemployment research. However, the quan-titative evaluation of the representative datasets that are currently available in the fieldof employment research reaches its limitswhen it comes to examining precisely whatinduces employees and employers to consentto overqualification; the present study couldonly provide some initial leads. A qualitativeapproach emerged here as an effective meansof evaluating the motivation of employees.

The findings of the only studies in which thistype of approach was adopted (Schlegelmilch,1982, 1983a, 1983b and 1987) cannot be gen-eralised because they are restricted to onelevel of qualification and because their sam-ple is unrepresentative and too small. Nonethe less, the approach displays an interest-ing analytical thrust, which is able to offerinsights that cannot be gleaned from the find-ings of conventional analyses; this makes it asuitable medium for imparting fresh momen-tum to quantitative research based on repre-sentative data by broadening the scope of sur-veys to include aspects such as the intrinsicmotivation of people to do jobs for which theyare overqualified. A first movement in thisdirection is made in Büchel, 1998b.

With regard to closer investigation of employ-ers’ motivation to recruit overqualified jobapplicants, Haugrund (1990) has been theonly author so far to highlight this specificresearch angle. However, the fact that hissurvey was limited to a few firms in a singleindustry and only examines technical staffwith fairly high skill levels means that hisfindings are not amenable to generalisationeither. More comprehensive future studieswith a broader database, designed to followup this preliminary work, could deliver im-portant new information.

7.2 Implications for education andemployment policies

7.2.1 Implications for education policy

The findings of this study give rise to differ-ent postulates for the domains of vocationaleducation and higher education, becausethese two parts of the education system nor-mally have widely different funding struc-

Felix Büchel

520

tures. The German Duales System of alter-nating practical and theoretical vocationaltraining of apprentices is discussed in some-what greater detail here, because it is oftenpresented as a model system by numerouspublic policymakers in the field of vocationaltraining in other European countries.

Dual vocational training

In relation to the German dual system,Neubäumer (1993) found that those indus-tries which experience difficulties in recruit-ing skilled workers tend to train in excess oftheir needs. These are generally industrieswith largely unattractive and undemandingjobs. Many of the trainees who are surplus torequirements have no option but to move tounderskilled jobs.

The findings of a study by the present author(Büchel, 1998b) confirm this effect, wherebyindividuals who have completed the basiclevel of secondary education and possess acertificate of apprenticeship run a dispropor-tionately high risk of overqualification. Thisshows that the quality of training in the Ger-man dual system is far from being as uniformas the advocates of this very characteristicallyGerman system of vocational training fre-quently assert and as might be suggested bythe annual discussion that takes place in latesummer regarding the situation in the ap-prenticeship market, a discussion which tendsto focus exclusively on the number of train-ing places available and the demand for thoseplaces.

The same findings prove, as did those previ-ously obtained by Hofbauer and Nagel (1987),for example, that the volume of skills andqualifications which the system produces butwhich the labour market does not use hasassumed menacing proportions, whether thismeans that the jobs to which those skills andqualifications relate are insufficiently attrac-tive or that there are simply no suitable jobsavailable. There is a need to identify the rea-sons for this inefficiency of the dual systemand to seek ways of improving it.

First of all, the perennial criticism of exces-sive specialisation in the vocational-training

system, reflected in a huge number of distincttrades, remains valid today. This specialisa-tion makes it more difficult for individuals touse their acquired skills if they subsequentlyhave to work outside the trade they havelearned. It would make better sense if empha-sis were placed on imparting selected keyskills, which should certainly be relevant togroups of specific occupations; specialisationshould be delayed until trainees have beguntheir careers, when the knowledge and skillsrequired for the performance of specific tasksshould be imparted through on-the-job train-ing, as is done in the United States. This pointhas now been taken on board by educationpolicymakers and initial moves have beenmade in the postulated direction.

In addition, training courses which are out ofdate and for which there is consequently lit-tle or no market demand should be discontin-ued, and at the same time courses should bedeveloped in new marketable fields of activ-ity. This process currently takes too long, notleast because of the strictly regulated proce-dures and the large number of interest groupsinvolved in decision-making. These problemsare now the subject of critical discussion inGermany, and some very promising propos-als have already emerged.

Another strategy focuses on the improvementof the existing information system regardingthe marketability of the various trainingcourses. The information provided by careersadvice centres is all too frequently confinedto a description of training curricula with nodetails of career prospects. The latter infor-mation should be provided before prospectivetrainees choose their courses. Hofbauer andNagel (1987) recommend that advice shouldbe available from the employment authoritiesin the final year of technical college (p.45),but this is clearly the wrong time. By thenthe fatal chain of events depicted by Neu-bäumer (1993) has already been set in mo-tion; apprentices training for problem occu-pations realise too late that they have madethe wrong choice.

If the information situation were improved,not only in terms of training curricula but alsoin terms of subsequent career opportunities,


521

in accordance with the strategy presented byChaberny and Schober (1982), this would al-ter the training decisions of school-leavers,thereby influencing the content and qualityof employers’ training programmes and of thetrainees’ subsequent jobs. It must be bornein mind, however, that the effectiveness ofsuch a strategy will diminish as the processof change on the demand side of the labourmarket gathers momentum.

The emphasis in the present study is onoverqualification among non-graduates. How-ever, the implications for this level of qualifi-cation cannot be identified in isolation, notleast because the general public and the aca-demic world closely associate mismatches inthe labour market with overqualified univer-sity graduates. For this reason, the followingparagraphs deal briefly with the implicationsof the overqualification problem for higher-education policy.

Higher education

In the domain of higher education, as in thedual system of vocational education and train-ing, the risk of overqualification is observablydependent on the profession for which the stu-dent is training or the discipline he or she isstudying (see for example Büchel and Ma-tiaske, 1996, and Alpin et al., 1998). Althoughin Germany university graduates in generalrun a somewhat lower risk of overquali-fication than holders of vocational diplomas,for example, once the data have been control-led for the main personal characteristics thedifferences become insignificant. Indeed,graduates of higher technical colleges(Fachhochschulen) – again after controllingfor personal characteristics – are significantlymore prone to overqualification than holdersof vocational diplomas (Büchel 1998b, Table5.8, p.204). There is also evidence to suggestthat the problem of overqualification will be-come more acute for German graduates – es-pecially women graduates – in the future (seeBüchel and Weißhuhn, 1997a and 1998, aswell as Büchel, 1996a).

The overqualification situation described inthis study gives rise to certain postulates withregard to education policy, but these cannot

be formulated without reference to the spe-cific discussion on graduate overqualification.The problems reflected in the findings thathave been obtained are subject to widely vary-ing interpretations, depending on the observ-er’s perspective. Nevertheless, the questionarises, at least in economic terms, as towhether the skills demanded by the labourmarket could not be provided by the educa-tion system with less effort and at less ex-pense. This question sheds light on three ofthe main reasons for the present glut of aca-demic qualifications.

First of all, the fact that higher education inthe vast majority of European countries islargely funded from the public purse createsprofound external effects. The excessivecheapness of university study by marketstandards systematically produces a surplusof qualifications; the introduction of tuitionfees, which are of mere symbolic value whencompared with the actual cost of universityeducation, will do little to change this.

Secondly, an academic degree may be re-garded as the most valuable formal qualifi-cation. According to the postulates underly-ing the job-competition model, it placesgraduates quite near the front of the labourqueue. This sort of position remains soughtafter, even when the returns to education andemployment prospects in general decline, forexample because of a glut of academic quali-fications or a drop in the standard of univer-sity education. The desire to secure the high-est possible educational qualification in adiscipline selected on the basis of personalpreference therefore remains rational from anindividual point of view, even in a climate ofdeteriorating career prospects (for a detailedtreatment of this aspect, see Zwick (forthcom-ing)); the fact that the aggregate macro-economic effect of such a strategy may be toreduce the general level of well-being is un-likely to have any bearing on the educationaldecisions that individuals make.79

79 On the competing aims of the manpower-require-ment approach and the social-demand approach,see Kühlewind and Tessaring, 1975.

Felix Büchel

522

Thirdly, the positive signal that is emitted bythe highest vocational qualification is alsovery desirable in terms of a person’s socialprofile, irrespective of employment status,and this also stimulates demand.

Demand for the top educational qualificationsmay therefore be expected to keep on risingas long as the educational attainment seemsto be ‘cheap’ in terms of the required personaleffort and financial outlay. This raises thequestion as to how education policy can beused to combat the resultant excess demandfor academic qualifications which is gradu-ally leading more and more university gradu-ates into overqualification.

A restrictive policy could begin with the in-troduction of tuition fees at market rates.These fees, however, would have to be basedon the actual cost of training, unlike the mod-els that are most often discussed at thepresent time. In terms of market economics,there is no logical reason why the same feeshould be payable for a (cheap) law course asfor a course in medicine, for example. In ad-dition, the intended effect of reducing the levelof demand for education can only be achievedin conjunction with an absolute repaymentobligation which must apply irrespective ofsubsequent career success, although provisionwould certainly have to be made for a mecha-nism designed to avoid social hardship. Theoverall effect of such linkage between the de-mand for education and self-financing wouldbe to reduce the size of the group of over-qualified employees who report that a success-ful career ‘is not so important’.

This, however, raises the problem of the socialacceptability of tuition fees at market rates,which, as the current public discussion hasshown, is a tough nut to crack. A more realis-tic means of restricting the demand for educa-tion would seem to be a reform of the univer-sity course and examination regulations basedon the modular acquisition of certificates (fora discussion of this option in the German con-text, see Büchel and Helberger, 1995).

As in the domain of non-graduate employ-ment, the proposed academic reforms wouldhave to be accompanied by a further improve-

ment of the system whereby prospective stu-dents can obtain information on career pros-pects for graduates of specific departmentsand institutions in each federal state; a newcertification system would also entail the pro-vision of information on the various types ofcertificate offered by each department andinstitution.80 This might prevent prospectivestudents from choosing courses in ‘high-risk’subjects, depending on the level of risk theywere prepared to contemplate and on whetherthe ‘low-risk’ option was overridden by theirflair for a particular subject, which would stillhave to be recognised as a legitimate selec-tion criterion. The studies on graduates inemployment that exist at the present time(see subsection 5.2 above) could serve as use-ful methodological guides, but the cumulativedata they contain are still far from constitut-ing the desired comprehensive informationmechanism.

7.2.2 Implications for employmentpolicy

Compared with the range of political mecha-nisms that are available in the educationalfield, the armoury of employment policies isless well equipped to improve the generalmatch between training levels and job re-quirements. The forms of direct interventionproposed by Tsang and Levin (1985, p.101),such as tax breaks for employers of appropri-ately qualified workers, hardly seem to bepracticable. A cautionary example of this typeof hapless external intervention in the labourmarket is the levy imposed on German com-panies which do not employ their statutoryquota of disabled workers. This measure hasfailed to achieve its intended aim of encour-aging the recruitment of disabled persons. Itseems likely that the introduction of a simi-lar levy on training places, which was underserious discussion in Germany until very re-cently, would have posed similar problems.

Measures designed to broaden the skills andqualifications of overqualified workers are

80 For a summary of the official system in Ger-many as it now stands, see Dostal et al., 1999,pp.57 ff.


523

likely to prove more effective.81 These meas-ures are generally targeted at unemployedjobseekers at the present time. As part of theprocess of reducing the specialised componentof curricula in the field of non-academic voca-tional training, consideration should be givento the possibility of extending such measuresto overqualified workers and to devising somemeans of dividing the cost between employ-ers and the government. Emphasis should beplaced on training in a marketable job, inother words within the framework of a properemployment contract. While the current wide-spread implementation of job-creationschemes in Germany fulfils an important so-cial function at the present time, when thereis no suitable alternative, it does distort themarket and means, for example, that the find-ings of conventional market studies, which arebased on employees’ qualification levels anddo not differentiate between workers on job-creation schemes and those in regular employ-ment, send erroneous signals to the educa-tion system about the utility of the certificatesit awards.

Limited regional mobility has proved to be anobstacle, at least in Germany, to people’s pros-pects of obtaining a job commensurate withtheir qualifications (Büchel, 1998b, pp.143 ff.).The relevant findings show that greater effi-ciency could be achieved not only if individu-als were willing to move house but even if theywere prepared to commute. All measures de-signed to overcome this reluctance to move toplaces where work is available, a reluctancethat has traditionally been far greater in Eu-rope than in the United States, for example,will help to ensure a better return on invest-ments in skills and qualifications and to en-hance the efficiency of the labour market.There are many conceivable strategies thattranscend the conventional mechanisms ofemployment policy, from structural measuressuch as improvement of the roads and trans-port infrastructure, including the develop-ment of local public transport services, to fis-

cal measures such as tax concessions to off-set work-related removal costs.

Here too, however, within the narrower con-fines of employment policy, an informationdeficit has to be highlighted. A long-distancemove in search of work presupposes knowl-edge of the employment opportunities at thenew location. Adequate information has hith-erto been reserved for the more highly quali-fied. The jobs that might interest them arenormally advertised nationally. By contrast,there is a severe shortage of information on anational, let alone European, scale forjobseekers with no managerial qualifications.Employers traditionally recruit these employ-ees locally. Given the identified benefits ofregional mobility, the establishment of a Eu-ropean recruitment service on the Internetwould certainly be a huge step forward.

The employment strategies described aboveare primarily designed to adjust the ratio ofmatches to mismatches. It is evident thatemployment policies also exert a strong in-fluence on the ratio of overqualified workersto jobseekers. An employment policy can beused to motivate skilled jobseekers to acceptjobs for which they are overqualified by cut-ting their unemployment benefit or by tight-ening the rules governing the definition of areasonable job offer and by ensuring that therules are enforced.

There are several grounds for suggesting thatthis sort of induced increase in overqualifi-cation at the expense of the unemploymentrate would be economically sound. In someEuropean countries, the percentage of low-paid jobs is judged by the OECD to be dispro-portionately small when compared with coun-tries such as the United States. This isprobably due to the fact that, depending onthe type and composition of the household,state benefit from the European welfare sys-tems, which are often corporatist in nature,may sometimes offer a higher income thanan unskilled or semi-skilled job.

On the other hand, trade unions are gradu-ally coming to realise that there is too muchstandardisation of wage rates at the lower endof the scale, which results in excessively high

81 The role of further education as a means of re-ducing the risk inherent in underemployment isdiscussed in Büchel, 1998b.

Felix Büchel

524

pay for unskilled and semi-skilled work. Whilegreater differentiation of wage rates at thelower end of the pay scale, accompanied by areduction in transfer payments, would cer-tainly increase the number of jobs – especiallyin the service sector, with its high job-crea-tion potential – we could also expect an in-crease in the numbers of the ‘working poor’,as has happened in the United States.

This strategy, however, can only operatewithin certain limits, which are set by socialpolicy or by society itself. For example, mostof the European population would surely feelrather uncomfortable with the idea that, asoften happens in the United States, everystore checkout would have an auxiliary – whomight be overqualified for the job – standingby to pack customers’ shopping into bags andcarry it to their cars, even though there mightbe a market for this service at a suitably lowrate of pay. The introduction of a combinedwage, made up of income from employmentand transfer payments, which is currentlyunder discussion, could be an effective meansof creating more jobs.

The present study has shown that significantlinks exist between overqualification and un-employment. So the question whether thepercentage of overqualification in a nationaleconomy is ‘too high’ or ‘too low’ must alwaysbe answered by reference to the unemploy-ment rate. The ideal ratio between theoverqualified and unemployed percentages ofthe labour force cannot be determined by ascientific study but needs to be establishednormatively in the light of society’s collectivevalues. The same applies to the percentage ofoverqualified workers when considered in iso-lation. Overqualification must not be seen asan entirely bad thing; it also has positive ef-fects, for example in the form of spillover ef-fects in connection with the skill upgradingof certain occupations.82 It is conceivable that,

like the rate of ‘natural’ unemployment, anoptimum level of overqualification could beidentified on the basis of structural models,although this too would surely spark off a dis-cussion on the considerations that should de-termine whether the labour market is func-tioning at maximum efficiency.

Summary

Section 1 begins by setting out the purpose ofthe study, which is to investigate the reasonsfor the occurrence of overqualification and toaddress the problem of finding a valid meansof identifying cases of overqualification. At thesame time, the author takes the innovativestep of trying to establish a typological affin-ity between overqualification and unemploy-ment (including the hidden labour reserve).This affinity derives on the one hand from thefact that both situations involve a mismatchbetween qualifications and job requirements.It may be assumed that both phenomena im-pose a heavy financial burden on nationaleconomies, since economies do not operate atfull capacity when part of their stock of hu-man capital lies dormant. This implies thatoverqualification is a problem that needs tobe addressed in the same way as unemploy-ment, although there can be no disputing thatunemployment is the more serious problem.In terms of the human capital theory, theanalogy may be seen as follows: in the case ofunemployment, 100% of a person’s acquiredformal vocational qualifications remain un-used; in the case of overqualification, x% ofthose qualifications are unused, x being a sig-nificant quantity which can certainly be meas-ured empirically, for instance by means of in-come analyses.

Another common feature of the two situationsis that they are potentially temporary in na-ture and that the vast majority of people inthem are not there out of choice. It may alsobe assumed that people in either situation willtend to aspire to a move into employment com-mensurate with their level of qualification.This means that it is possible in principle toanalyse the phenomenon of overqualificationwith the same dynamic research strategy thatwas developed in the seventies to study un-

82 On this point, see the study by Planas et al. inthe present volume. However, the questionwhether the skill upgrading referred to here or acompetitive ‘bumping down’ is the dominant proc-ess is essentially determined by the elasticity ofmarket demand for higher qualifications (for moredetails, see Borghans and de Grip, 1999).


525

employment. This standpoint, however, is stillvery rarely encountered in the field of employ-ment research.

Section 2 deals with terminological matters.Even at this stage, overqualification researchis complicated by an unwieldy mass of con-cepts.

Section 3 briefly presents the various theo-ries with which researchers have tried to ex-plain the persistence of overqualification inthe labour market. Some unorthodox ap-proaches are also outlined. Most of the theo-ries can only claim to explain part of the phe-nomenon. This section also discusses sets ofinfluential factors that have scarcely everbeen treated in the existing body of literature:institutional conditions such as the rules gov-erning the system of unemployment benefit,basic economic conditions such as the tradecycle or the situation in the regional labourmarket, the personal preferences of those whooffer their labour, in terms of the utility theyascribe to career success, etc., and the pro-ductivity calculations of employers, who needto know whether overqualified workers areliable to outperform their less-qualified col-leagues.

In section 4, the problem of identifying in-stances of overqualification is systematicallydiscussed. The discussion starts with an ex-planation of the two main measurement strat-egies – the ‘objective’ and the ‘subjective’ ap-proach – and their diverse variants; thepresentation includes unorthodox measure-ment processes.

The so-called ‘objective’ approach identifiesoverqualification by comparing the employ-ees’ formal qualification levels with the re-ported information concerning their occupa-tions. This approach depends on preciseknowledge of the training that exists for par-ticular occupations. Changes in the level ofoccupational requirements in the course oftime, a high degree of variation between jobrequirements within the same occupationalcategory and the typically high number ofmissing data on occupations that results fromthe conversion into ISCO codes of occupationsthat respondents are normally asked to re-

port in words all pose considerable problemswhen this approach is adopted.

The so-called ‘subjective’ approach, on theother hand, compares the employees’ formalqualification levels with their own subjectiveassessments of the qualifications required fortheir respective jobs. The disadvantage hereis undoubtedly the greater element of uncer-tainty that always accompanies subjectiveassessments. A major advantage of this ap-proach, however, lies in the assumption thatthe employees themselves are the best peo-ple to ask about the real requirements of theirjobs. The methodological discussion that waslargely conducted in U.S. research journalsduring the eighties essentially came to theconclusion that the subjective approach pos-sessed greater validity than the objective ap-proach. Accordingly, research designs basedon the objective approach are hardly ever en-countered in contemporary studies.

Once these strategies have been explained,the author’s own refinement of the conven-tional subjective measurement strategy isdescribed. This is principally based on theintroduction of a third variable – occupationalstatus – into the categorisation process. Thisvariable serves to validate the standard com-parison between the formal qualification andthe employee’s subjectively reported job-re-quirement level. The overqualification vari-able created by this measurement process,which still takes the dichotomous form‘overqualified? yes/no’, turns out to be consid-erably more selective than the variable gen-erated by the conventional two-variable proc-ess.

The validity of the construct was tested withthe aid of job characteristics (about 50 items)of overqualified and adequately employedworkers; differences – usually substantial –between these two categories were observedfor almost every item. The only disadvantageis the emergence of two additional categories.The first of these covers cases that cannot bedefinitively assigned to either of the originalcategories, because the combination of thethree input variables does not permit a clearclassification as overqualification or adequateemployment, while the second category cov-

Felix Büchel

526

ers cases where the combination of the threevariable characteristics is implausible. Thesetwo categories are generally excluded from theempirical evaluation. The reduction in thetotal number of cases, however, is fairly small.

Section 5 provides a comprehensive review ofthe literature that deals with the subject ofoverqualification. Special emphasis is placedon the origins of this discussion in the wake ofthe expansion of education systems in the sev-enties and eighties. The beginnings of the aca-demic investigation of the problem of over-qualification are retraced for the United Statesfirst of all. This is the country from which themain contributions to this field of researchhave come, contributions which, above all else,set the methodological standard.

The early development of overqualificationresearch outside the United States is de-scribed, the German situation being used asan example. Thereafter, more recent worksfrom the United States and Europe – includ-ing international comparative studies – arepresented. It emerges that European over-qualification researchers have only begun inrecent times to draw upon the methodologicalgroundwork that was produced in the UnitedStates. It is also noticeable that, apart fromthe Dutch, British and German research, thework that has been done on overqualificationin Europe is still at a very rudimentary stage.

In section 6, on the basis of German longitu-dinal data, some selected examples of issuesin the field of overqualification research areexamined empirically in the light of the pos-tulated affinity between unemployment andoverqualification.

It emerges first of all from cross-section analy-ses that, once the data have been controlledfor the main social characteristics, over-qualified workers do not differ significantlyfrom jobseekers in their problem-solving ca-pacity and level of political involvement butthat both lag far behind workers whose jobsmatch their qualifications. In terms of moraleand concerns about the future, the over-qualified group is positioned halfway betweenthe unemployed and adequately employedgroups.

The second stage of the examination involvesthe use of the instruments of dynamic unem-ployment research to analyse overquali-fication. Firstly, transitions into overquali-fication from various other types ofemployment status are examined. Themultivariate analyses show that, when thedata are controlled for the respondents’ so-cioeconomic background, unemployment isthe status from which individuals most fre-quently move into overqualification. This isevidence of important exchange processesbetween these two types of employment sta-tus. Transitions from overqualification to un-employment (downward mobility) and to edu-cationally adequate employment (upwardmobility) are then analysed.

These analyses demonstrate that transitionsinto unemployment occur more frequentlythan transitions to adequate employment.This finding, which is consistent with the pos-tulates of the segmentation approach, can beseen as further proof of exchange process be-tween overqualification and unemployment.In an additional analysis, the study examineswhether a phase of overqualification follow-ing a period of adequate employment is likelyto afford protection against unemployment orwhether it tends to be one step in a down-ward process. The former is the case, whichindicates that employees with a strong desireto work who see their educationally adequatejob at risk or who have already been maderedundant will protect themselves againstunemployment if they take a flexible approachand accept a second-best option in the formof overqualification.

A transition into adequate employment, how-ever, is a more frequently observable occur-rence among unemployed than overqualifiedindividuals. An additional analysis also dem-onstrates that acceptance of an underskilledjob does not bring jobseekers any closer toadequate employment. On the one hand, thismay mean that many underskilled jobs canbe regarded as dead-end jobs. On the otherhand, this finding could indicate that a sig-nificant number of unemployed people arepursuing an inflexible job-search strategy andwill only accept jobs commensurate with theirqualifications. This possibility touches on the


527

question of the institutional rules governingunemployment benefit.

Lastly, longer-term income effects of over-qualification are examined. It emerges thatthe incomes of overqualified workers growmore slowly than those of their adequatelyemployed counterparts. This remains the caseeven after the basic income level has beencontrolled for numerous socioeconomic char-acteristics. A final additional analysis is thenconducted to establish whether – as would beexpected – a period spent in overqualificationcreates income effects that lie somewherebetween those of adequate employment andthose of unemployment. The expectations ofthe human capital theory are indeed fulfilled:if periods of overqualification or unemploy-ment intervene in the course of a career spentmainly in adequate employment, while thephases of overqualification generate signifi-cant income losses by comparison with un-broken adequate employment, these lossesare lower than those that result from a pe-riod of unemployment.

The study is rounded off in section 7 withmethodological conclusions addressed toemployment researchers and substantive im-plications for education and employment poli-cies. The first postulate addressed to employ-ment researchers is that the typologicalaffinity between overqualification and unem-ployment should be recognised; this wouldresult in far greater diversity of researchstrategies for the analysis of overqualification,particularly longitudinal analysis, than hashitherto been observable in the relevant lit-

erature. The second postulate is that predic-tions of future demand for labour should nolonger be based, as has hitherto been custom-ary, on a prolongation of the development ofan employee’s level of qualification but ratheron the development of the skill level requiredfor his or her job. Neglecting the aspect ofmismatch by disregarding the problem ofoverqualification could lead to glaring errorsin the assessment of future labour require-ments. The likelihood of such miscalculationwill be greater if overqualification rates riseover a period of time.

A host of suggestions are made topolicymakers in the fields of education andemployment as to how the percentage ofoverqualified workers in the labour force canbe reduced. These are largely extracted bymeans of partial analysis from the empiricalfindings set out in the available literature.For details of these suggestions, the reader isreferred to the short subsection 7.2 of thestudy. Here too, the crux of the matter is thatthe problem of overqualification should beconsidered relevant in macroeconomic termsand that political pressure should be appliedwith a view to solving it; such pressure shouldbe structured in the same way as the mecha-nisms used to combat unemployment, al-though lower priority will naturally be givento overqualification. Both overqualificationand unemployment are signs of inadequatecoordination within the labour market, inwhich the inefficiencies of the education sys-tem play a significant role. Accordingly, bothproblems lend themselves to treatment withsimilar sets of mechanisms.

Felix Büchel

528

Adamy W., Bosch G., 1990. Arbeitsmarktprojektion– Wegweiser für Beschäftigungspolitik?Mitteilungen aus der Arbeitsmarkt- undBerufsforschung, 23(1), pp.111-121.

Alba-Ramírez A., 1993. Mismatch in the Spanishlabor market: Overeducation? The Journalof Human Resources, 28(2), 1993, pp. 259-278.

Alpin C., Shackleton JR., Walsh S., 1998. Over-and Under-Education in the UK GraduateLabour Market. Studies in Higher Educa-tion, 23(1), pp. 17-34.

Amick B.C., Lavis J.N., 1998. Labor Markets andHealth: A Framework and Set of Applica-tions. Paper submitted to the Kansas Con-ference on Health and Its Determinants,Wichita, Kansas, April 20-21.

Amirault T.A., 1992. Job Market Profile of CollegeGraduates in 1992: A Focus on Earnings andJobs. Occupational Outlook Quarterly, 20(Summer), pp.20-28.

Arrow K.J., 1973. Higher Education as a Filter.Journal of Public Economics, 2, pp.193-216.

Batenburg R., Witte M. de, 1998. Underemploy-ment in the Netherlands. How the Dutch‘poldermodel’ failed to close the education-job gap (submitted manuscript).

Battu H., Belfield C. R., Sloane P.J., 1999. Over-education among Graduates: a Cohort View.Educational Economics, 17(1), pp.1-18.

Battu H., Seaman P., Sloane P., 1998a. Are Mar-ried Women Spatially Constrained? A Testof Gender Differentials in Labour MarketOutcomes. Discussion Paper 98-07, Depart-ment of Economics, University of Aberdeen,UK.

Battu H., Seaman P., Sloane P., 1998b. The Im-pact of Regional Migration on the Earnings,Employment and Overeducation of MarriedWomen (submitted manuscript).

Battu H., Sloane P., forthcoming. Overeducation andCrowding out in Britain. In: L. Borghans, A.deGrip, eds. The Overeducated Worker? TheEconomics of Skill Utilization. Edward Elgar,Cheltenham, UK, and Northampton, Mass.,USA.

Becker G.S., 1964. Human Capital. ColumbiaUniversity Press, New York.

Beneito P., Ferri J., Molto M.L., Uriel E., forth-coming. Over/Undereducation and SpecificTraining in Spain: Complementary or Sub-stitute Components of Human Capital? In:H. Heijke, J. Muysken, eds. Education andTraining in a Knowledge-based Economy.AEA-Macmillan Series, Macmillan Press,Basingstoke, UK.

Berg I., 1970. Education and Jobs: The GreatTraining Robbery. Praeger, New York.

Berg I., 1989. Rich Man´s Qualification for PoorMan´s Jobs. Trans-Action, 6(5), pp.45-50.

Bielenski H., Brinkmann C., Kohler B., 1994.Erwerbsverläufe und arbeitsmarktpolitischeMaßnahmen in Ostdeutschland. Ergebnissedes Arbeitsmarkt-Monitors über beruflicheVeränderungen 1989 bis 1993. IAB workshopreport No 12.

Bielenski H., Brinkmann C., Kohler B., 1995.Erwerbsverläufe seit der Wende in Ost-deutschland. Ergebnisse des Arbeitsmarkt-Monitors über berufliche Veränderungen1989 bis 1994. IAB workshop report No 6.

Bishop J., 1995. Overeducation. In: M. Carnoy, ed.International Encyclopedia of Economics ofEducation. 2nd Edition. Pergamon Press,Oxford, UK, pp. 375-381.

Blaschke B., 1987. Soziale Qualifikationen im Er-werbsleben – Theoretisches Konzept und em-pirische Ergebnisse. Beiträge zur Arbeits-markt- und Berufsforschung 116, Nuremberg.

Blechinger D., Pfeiffer F., 1999. Qualifikation,Beschäftigung und technischer Fortschritt.Empirische Evidenz mit den Daten desMannheimer Innovationspanels. Jahrbü-cher für Nationalökonomie und Statistik,218(1/2), pp.129-145.

Blien U., 1986. Unternehmensverhalten undArbeitsmarktstruktur. Eine Systematik undKritik wichtiger Beiträge zur Arbeitsmarkt-theorie. Beiträge zur Arbeitsmarkt- undBerufsforschung, Vol. 103. Institut für Ar-beitsmarkt- und Berufsforschung (IAB),Nuremberg.

Bodenhöfer H.-J., 1984. Unternehmenswachstum,Qualifikationsstruktur und Absorption eineshöheren Qualifikationsangebots. In: H.P.Widmaier, ed. Das Arbeitskräfteangebot zwi-schen Markt und Plan. Duncker & Humblot,Berlin, 1984, pp.11-60.

Bibliography


529

Borghans L., Bruinshoofd A., Grip A. d., (forthcom-ing). Low Wages, Skills and the Utilizationof Skills. In: L. Borghans, A. deGrip, eds.The Overeducated Worker? The Economicsof Skill Utilization. Edward Elgar, Chelten-ham, UK, and Northampton, Mass., USA.

Borghans L., Grip A. d., 1999. Skills and low pay:upgrading or overeducation? (submittedmanuscript). ROA, Maastricht University.

Borghans L., Grip A. d., forthcoming(a). The De-bate in Economics about Skill Utilization.In: L. Borghans, A. deGrip, eds. The Over-educated Worker? The Economics of SkillUtilization. Edward Elgar, Cheltenham, UK,and Northampton, Mass., USA.

Borghans L., Grip A. d., ed., forthcoming(b). TheOvereducated Worker? The Economics ofSkill Utilization. Edward Elgar, Chelten-ham, UK, and Northampton, Mass., USA.

Borghans L., Grip A. d., Sloane P., 1998. Under-utilisation of skills, bumping down and lowwages (submitted manuscript).

Borghans L., Smits W., 1997. Underutilisation andwages of HVE graduates. Paper submittedto the AEA Conference, Maastricht, NL.

Bowles S., Gintis H., 1976. Schooling in Capital-ist America. Basic Books, New York.

Bowlus A. J., 1995. Matching Workers and Jobs:Cyclical Fluctuations in Match Quality. Jour-nal of Labor Economics, 13(2), pp. 335-350.

Brinkmann C., Wiedemann E., 1995. Arbeits-marktrisiken im ostdeutschen Transforma-tionsprozeß: Ergebnisse des Arbeitsmarkt-Monitors 1989 bis 1994. Mitteilungen ausder Arbeitsmarkt- und Berufsforschung,28(3), pp.323-338.

Brinkmann C., Wiedemann E., 1997. UnterwertigeBeschäftigung. In: H. Bielenski, ed. DerArbeitsmarkt Ostdeutschlands im Umbruch:Datensätze, Methoden und ausgewählteErgebnisse des Arbeitmarkt-Monitors 1989– 1994. Beiträge zur Arbeitsmarkt- undBerufsforschung 210, IAB, Nuremberg,pp.56-60.

Brinkmann G., 1978. Die Absorption der Hoch-schulabsolventen durch das Beschäftigungs-system der Vereinigten Staaten von Ameri-ka. Mitteilungen aus der Arbeitsmarkt- undBerufsforschung, 11(4), pp. 425-434.

Büchel F., 1992. Die Qualität der Wiederbeschäfti-gung nach ununterbrochener und nach «per-forierter» Langzeitarbeitslosigkeit. Beiträgezur Arbeitsmarkt- und Berufsforschung 162,IAB, Nuremberg.

Büchel F., 1994a. Die Stabilisierung extreminstabiler Beschäftigungsverläufe. Einesegmentations-theoretische Betrachtung.Jahrbücher für Nationalökonomie undStatistik, 213 (4), pp. 456-470.

Büchel F., 1994b. Overqualification at the Begin-ning of a Non Academic Working Career. TheEfficiency of the German Dual System un-der Test. In: G.-A. Horn, H. Trabold, eds.Globalisation and Structural Unemploy-ment. Special Issue of Konjunkturpolitik40(3-4). Duncker & Humblot, Berlin, pp.342-368.

Büchel F., 1995. Unterwertige Beschäftigung inOstdeutschland. Erste Informationen zuGrößenordnungen und Entwicklungs-tendenzen. Empirische Arbeitsmarkt-forschung zur Transformation in Ost-deutschland. In: D. Beer, C. Brinkmann, A.Deeke, S. Schenk, eds. Reihe «SAMF-Ar-beitspapiere», Band 1995-4. SAMF, Gelsen-kirchen, pp.187-195.

Büchel F., 1996a. Der hohe Anteil an unterwertigBeschäftigten bei jüngeren Akademikern:Karrierezeitpunkt- oder Strukturwandel-Effekt? Mitteilungen aus der Arbeitsmarkt-und Berufsforschung, 29(2), pp.279-294.

Büchel F., 1996b. Wenn sich der Sozialpädagogeals Kellner durchschlägt. Guest article forthe education page of the FrankfurterRundschau of 19 September 1996, p.6.

Büchel F., 1997. Berufseinmündung und ersteKarrierephase von Akademikern mit einerberuflichen Zusatzqualifikation. Ein Beitragzur Kontroverse ‘HIS versus Büchel/Helberger’. Mitteilungen aus der Arbeits-markt- und Berufsforschung, 30(3), pp.620-634.

Büchel F., 1998a. Unterwertig Erwerbstätige –Eine von der amtlichen Statistik über-sehene Problemgruppe des Arbeitsmarktes’,In: J. Schupp, F. Büchel, M. Diewald, R.Habich, eds. Arbeitsmarktstatistik zwischenRealität und Fiktion. ‘Sozialstruktur undSozialberichterstattung’ series of the BerlinSocial Science Research Centre (WZB).edition sigma, Berlin, pp.113-130.

Felix Büchel

530

Büchel F., 1998b, Zuviel gelernt? Ausbildungs-inadäquate Erwerbstätigkeit in Deutsch-land. W.Bertelsmann Verlag, Bielefeld.

Büchel F., 1999a. The Effects of Overeducation onProductivity in Germany – The Firms’ View-point. Paper presented at the EALE 1999Conference, Regensburg (Germany), 23-26September 1999.

Büchel F., 1999b. Productivity Effects of Overedu-cation in Germany – First Results. In: T.A.Dunn, J.R. Frick, J.C. Witte, eds. Proceed-ings of the 1998 Third International Con-ference of the German Socio-Economic Pan-els Study Users. Special issue of theVierteljahrshefte zur Wirtschaftsforschung,68(2). Duncker & Humblot, Berlin, pp.199-203.

Büchel F., 1999c. Produktivitätseffekte durchÜberqualifikation. In: R.K. von Weizsäcker,ed. Bildung und Beschäftigung. Documentsof the Verein für Socialpolitik. Duncker &Humblot, Berlin.

Büchel F., 1999d. The Theory of ‘DifferentialOverqualification’ – Does it Work?

Büchel F., forthcoming. Tied Movers, Tied Stayers– The Higher Risk of Overeducation AmongMarried Women in West Germany. In: S.Gustafsson, D. Meulders, eds. Gender andthe Labour Market. Econometric Evidence onObstacles in Achieving Gender Equality.AEA-Macmillan Series, Macmillan Press,Basingstoke, UK.

Büchel F., Falter J. W., 1994. Der Einfluß von Lang-zeitarbeitslosigkeit auf die Parteibindung inder Bundesrepublik Deutschland. Zeitschriftfür Parlamentsfragen, 25(2), pp.186-202.

Büchel F., Helberger C., 1995. Bildungsnachfrageals Versicherungsstrategie. Der Effekt eineszusätzlich erworbenen Lehrabschlusses aufdie beruflichen Startchancen von Hoch-schulabsolventen. Mitteilungen aus der Ar-beitsmarkt- und Berufsforschung, 28(1),pp.32-42.

Büchel F., Matiaske W., 1996. Die Ausbildungs-adäquanz der Beschäftigung bei Berufsan-fängern mit Fachhoch- und Hochschul-abschluß. Konjunkturpolitik, 42(1), pp.53-83.

Büchel F., Mertens A., 2000. Overeducation,Undereducation, and the Theory of CareerMobility Paper presented at the EALE/SOLE 2000 Conference, Milan (Italy), 22-25 June 2000.

Büchel F., Mertens A., forthcoming. BeruflicheAufstiegschancen von Über- und Unter-qualifizierten – Ein Test der Karriere-mobilitäts-Theorie. In: U. Backes-Gellner, P.Moog, eds. Bildungssystem und betrieblicheBeschäftigungsstrategien. Documents of theVerein für Socialpolitik. Duncker &Humblot, Berlin.

Büchel F., Pannenberg M., 1997. Bildung. In: Sta-tistisches Bundesamt, ed. Datenreport 1997– Zahlen und Fakten über die Bundesrepu-blik Deutschland. Documents of the FederalCentre for Political Education (BpB),Vol.340. BpB, Bonn, pp.481-488.

Büchel F., Weißhuhn G., Bildungswege und Berufs-eintritt im Wandel. Mittelfristige Entwick-lung und sozio-ökonomische Bestimmungs-faktoren der Bildungsnachfrage und derÜbergangsmuster zwischen beruflichen Aus-bildungsformen in Deutschland. W. Ber-telsmann Verlag, Bielefeld, 1995.

Büchel F., Weißhuhn G., 1997a. Ausbildungs-inadäquate Beschäftigung der Absolventendes Bildungssystems. Berichterstattung zuStruktur und Entwicklung unterwertigerBeschäftigung in West- und Ostdeutschland.Report commissioned by the Federal Mi-nistry of Education, Science, Research andTechnology, Bonn. Volume 471 of the Volks-wirtschaftliche Schriften series. Duncker &Humblot, Berlin.

Büchel F., Weißhuhn G., 1997b. Ausbildungs-inadäquate Beschäftigung in Deutschlandund den USA. Ein Vergleich von Struktur undEinkommenseffekten auf der Basis vonPaneldaten. Report commissioned by the Fed-eral Ministry of Education, Science, Researchand Technology (BMBF), Bonn. BMBF, Bonn.

Büchel F., Weißhuhn G., 1997c. Unter Wert verkauft– Ausbildungsinadäquate Beschäftigungvon Frauen in West- und Ostdeutschland.Arbeitsbedingungen und sozioökonomischeBestimmungsgründe. Report commissionedby the Federal Ministry of Education,Science, Research and Technology, Bonn. W.Bertelsmann Verlag, Bielefeld.

Büchel F., Weißhuhn G., 1998. Ausbildungs-inadäquate Beschäftigung der Absolventendes Bildungssystems II. Fortsetzung der Be-richterstattung zu Struktur und Entwick-lung unterwertiger Beschäftigung in West-und Ostdeutschland (1993 – 1995). Reportcommissioned by the Federal Ministry ofEducation, Science, Research and Techno-logy, Bonn. Volume 471/II of the Volkswirt-schaftliche Schriften series. Duncker &Humblot, Berlin.


531

Büchel F., Witte J.C., 1997. The Incidence andConsequences of Overeducation AmongYoung Workers in the United States andGermany: A Comparative Panel Analysis.In: T. A. Dunn, J. Schwarze, eds. Proceed-ings of the 1996 Second International Con-ference of the German Socio-Economic Pan-els Study Users. Special issue of Viertel-jahrshefte zur Wirtschaftsforschung, 66(1),pp. 32-40.

Büchtemann C. F., Schupp J., Soloff D.J., 1993.Übergänge von der Schule in den Beruf –Deutschland und USA im Vergleich. Mittei-lungen aus der Arbeitsmarkt- und Berufs-forschung, 26(4), pp.507-520.

Burdett K., 1978. A Theory of Employee Searchand Quits. American Economic Review, 68,pp.212-220.

Burris B.H., 1983a. The Human Effects of Under-employment. Social Problems, 31, pp. 96-110.

Burris B.H., 1983b. No Room at the Top, Praeger,New York.

Burris V., 1983c. The Social and Political Conse-quences of Overeducation. American Socio-logical Review, 48(4), pp.454-467.

Busch D.W., Hommerich C., 1983. Der akademi-sche ‘Modellathlet’: Tendenzen, Erwartun-gen, Widersprüche, Hochschulexpansionund Arbeitsmarkt. Problemstellungen undForschungsperspektiven. In: ArbeitskreisBerufsforschung, Bremen and Bergisch-Gladbach, Institut für Arbeitsmarkt- undBerufsforschung, Nuremberg, Max-Planck-Institut für Bildungsforschung, Berlin, andWissenschaftliches Zentrum für Berufs- undHochschulforschung, Kassel, eds. Hoch-schulexpansion und Arbeitsmarkt. Problem-stellungen und Forschungsperspektiven.Beiträge zur Arbeitsmarkt- und Berufs-forschung 77. IAB, Nuremberg, pp. 60-87.

Buttler F., Tessaring M., 1993. Humankapital alsStandortfaktor. Argumente zur Bildungs-diskussion aus arbeitsmarktpolitischerSicht. Mitteilungen aus der Arbeitsmarkt-und Berufsforschung, 26(4), pp. 467-476.

Cain G., 1976. The challenge of segmented labormarket theories to orthodox theory: a sur-vey. Journal of Economic Literature, 14(3),pp.827-855.

Caplan R., Cobb S., French J., Harrison R.,Pinneau S., 1980. Job Demands and WorkerHealth. Institute for Social Research, AnnArbor, Mich., USA.

Cappelli P., 1993. Are Skill Requirements Rising?Evidence From Production and ClericalJobs. Industrial and Labor Relations Re-view, 46(3), pp.515-530.

Carnegie Commission on Higher Education, 1973.College Graduates and Jobs: Adjusting to aNew Labor Market Situation. McGraw-Hill,New York.

Cedefop, Tessaring M., 1998. Training for a chang-ing society: A report on current vocationaleducation and training research in Europe.Office for Official Publications of the Euro-pean Communities, Luxembourg.

Chaberny A., Parmentier K., Schnur P., 1982.Ausbildungsspezifische Strukturdaten. Teil1: Betriebliche Ausbildung. Beiträge zurArbeitsmarkt und Berufsforschung 61.1.IAB, Nuremberg.

Chaberny A., Schober K., 1982. Risiko und Chan-ce bei der Ausbildungs- und Berufswahl –Das Konzept der differenzierten Informati-on über Beschäftigungsaussichten. In: D.Mertens, ed. Konzepte der Arbeitsmarkt- undBerufsforschung. Eine Forschungsinventurdes IAB. Beiträge zur Arbeitsmarkt- undBerufsforschung 70. IAB, Nuremberg.

Clogg C.C., 1979. Measuring Underemployment:Demographic Indicators for the UnitedStates. Acadamic Press, New York.

Clogg C.C., Shockey J.W., 1984. Mismatch betweenOccupation and Schooling: A PrevalenceMeasure, Recent Trends and DemographicAnalysis. Demography, 21(2), pp.235-257.

Clogg C.C., Shockey J.W., 1985. The effect of chang-ing demographic composition on recenttrends in underemployment. Demography,22, pp.395-414.

Clogg C.C., Sullivan T.A., 1983. Labor Force Com-position and Underemployment Trends,1969-1980. Social Indicators Research, 12,pp.117-152.

Clogg C.C., Sullivan T.A., Mutchler J.E., 1986.Measuring Underemployment and Inequal-ity in the Work Force. Social Indicators Re-search, 18, pp.375-393.

Coburn D., 1975. Job-worker incongruence: con-sequence for health. Journal of Health andSocial Behavior, 16, pp.198-212.

Cohen M.S., 1971. Sex Differences in Compensa-tion. The Journal of Human Resources, 6,pp.434-447.

Cohn E., 1992. The Impact of Surplus Schoolingon Earnings: Comment. The Journal of Hu-man Resources, 27(4), pp.679-682.

Felix Büchel

532

Cohn E., Khan S., 1995. The wage effects ofoverschooling revisited. Labour Economics,2(1), pp.67-76.

Cohn E., Ng Y.C., 1999. Incidence and Wage Ef-fects of Overschooling and Underschoolingin Hong Kong. Economic Working Paper Se-ries, No. B-99-1, College of Business Admin-istration, University of South Carolina,USA.

Cohn E., Ng Y.C., Khan S., 1999. The Incidence ofOverschooling and Underschooling and itsEffect on Earnings in the United States andHong Kong (submitted manuscript).

Cornwell C., Rupert P., 1997. Unobservable Indi-vidual Effects, Marriage and the Earningsof Young Men. Economic Inquiry, 35, pp.285-294.

Daly M.C., Büchel F., Duncan G.J., 1997. Premi-ums and Penalties for Over- and Under-education: Cross-time and Cross-nationalComparisons in the United States and Ger-many. In: T.A. Dunn, J. Schwarze, eds. Pro-ceedings of the 1996 Second InternationalConference of the German Socio-EconomicPanels Study Users. Special issue of Viertel-jahrshefte zur Wirtschaftsforschung, 66(1).Duncker & Humblot, Berlin, pp.25-31.

Daly M.C, Büchel F., Duncan G.J., forthcoming.Premiums and Penalties for Over- andUndereducation: Cross-time and Cross-na-tional Comparisons in the United States andGermany. In: H. Oosterbeek, ed. Economicsof Education Review, special issue onovereducation.

Denison E., 1979. Accounting for Slower EconomicGrowth. Brookings Institution Press,Wahington, D.C.

Der Spiegel, Akademiker: «Schotten dicht». Arti-cle (no author, no issue number) in DerSpiegel magazine, 1985.

Doeringer P.B., Piore M.J., 1971. Internal LaborMarkets and Manpower Analysis. HeathLexington Books, Lexington, Mass., USA.

Dolton P., Vignoles A., forthcoming. The Incidenceand Effect of Overeducation in the UK Gradu-ate Labour Market. In: H. Oosterbeek, ed.Economics of Education Review, special is-sue on overeducation.

Dolton P.J., Vignoles A.F., 1997. OvereducationDuration. How long did graduates in the

1980s take to get a graduate level job? (sub-mitted manuscript).

Dooley M.D., 1986. The overeducated Canadian?Changes in the relationship among earn-ings, education, and age for Canadian men:1971-81. Canadian Journal of Economics,19(1), pp.142-159.

Dore R.P., 1976. The Diploma Disease: Education,Qualification and Development. Allen &Unwin, London.

Dostal W., Parmentier K., Schade H.-J., 1999.Möglichkeiten und Grenzen der quantitati-ven Berufsforschung im IAB – Eine Be-standsaufnahme. Mitteilungen aus der Ar-beitsmarkt- und Berufsforschung, 32(1),pp.41-60.

Duncan G.J., Hoffmann S.D., 1978. The economicvalue of surplus education. In: G. and M.Duncan, N. James, eds., Five ThousandAmerican Families. Vol. VI. Institute for So-cial Research, AnnArbor, Mich., USA,pp.233-246.

Duncan G.J., Hoffmann S.D., 1981. The Incidenceand Wage Effects of Overeducation. Econom-ics of Education Review, 1(1), pp.75-86.

Eckaus R.S., 1964. Economic Criteria for Educa-tion and Training. Review of Economics andStatistics, 46, pp.181-190.

Eijs P. van, Heijke H., forthcoming. Mismatch be-tween Occupation and Education and theCosts and Benefits of Job-related Training.In: H. Heijke, J. Muysken, eds. Educationand Training in a Knowledge-based Economy.AEA-Macmillan Series, Macmillan Press,Basingstoke, UK.

Fine S.A., 1968. The Use of the Dictionary of Oc-cupational Titles as a Source of Estimatesof Educational and Training Requirements.Journal of Human Resources, 3, pp.363-375.

Forgeot G., Gautié J., 1997a. Insertion profes-sionelle des jeunes et procéssus de déclasse-ment. Economie et statistique, 4/5 (304/305),pp.53-74.

Forgeot G., Gautié J., 1997b. Overeducation in theYouth Labour Market in France. Paper sub-mitted to the AEA Conference, Maastricht,NL.

Frank R.H., 1978a. Family Location Constraintsand the Geographic Distribution of Female


533

Professionals. Journal of Political Economy,86(1), pp.117-130.

Frank R.H., 1978b. Why Women Earn Less: TheTheory and Estimation of DifferentialOverqualification. American Economic Re-view, 68(3), pp.360-373.

Franz W., 1991. Arbeitsmarktökonomik. Springer,Berlin, etc.

Freeman R.B., 1971. The Market for College-Trained Manpower: A Study in the Econom-ics of Career Choice. Harvard UniversityPress, Cambridge, Mass., USA.

Freeman R.B., 1975a. Legal ‘cobwebs’: A recursivemodel of the market for new lawyers. Re-view of Economics and Statistics, 57, pp. 171-179.

Freeman R.B., 1975b. Overinvestment in CollegeTraining? Journal of Human Resources,10(3), pp.187-309.

Freeman R.B., 1975c. Supply and salary adjust-ments to the changing science manpowermarket: Physics. American Economic Re-view, 65(1), pp.27-39.

Freeman R.B., 1976a. A cobweb model of the sup-ply and starting salary of new engineers.Industrial and Labor Relations Review,29(2), pp.236-248.

Freeman R.B., 1976b. The Overeducated Ameri-can. Academic Press, New York, etc.

Freeman R.B., 1977. The decline in the economicrewards to college education. Review of Eco-nomics and Statistics, 59, pp.18-29.

Freeman R.B., 1980. The Facts About the Declin-ing Economic Value of College. The Journalof Human Resources, 15(1), pp.124-142.

Freeman R.B., 1981a. Career patterns of collegegraduates in a declining job market. NBERWorking Papers No 697.

Freeman R.B., 1981b. Implications of the chang-ing U.S. labor market for higher education.NBER Working Papers No 750.

Freeman R.B., Breneman D., 1974. Forecasting thePh.D. Labor Market. Technical Report 2,National Board of Graduate Education,Washington, D.C.

Fuchs V.R., 1971. Differences in Hourly EarningsBetween Men and Women. Monthly LaborReview, 94, pp.9-15.

Gautier P., forthcoming. Do more High-skilledWorkers occupy Simple Jobs during BadTimes? In: L. Borghans, A. deGrip, eds. TheOvereducated Worker? The Economics ofSkill Utilization. Edward Elgar, Chelten-ham, UK, and Northampton, Mass., USA.

Gill A.M., Solberg E.J., 1992. Surplus Schoolingand Earnings: a Critique. The Journal ofHuman Resources, 27(4), pp.683-689.

Gleiser S., 1996. Der Arbeitsmarkt für Akademi-ker. In: M. Tessaring, ed. Die Zukunft derAkademikerbeschäftigung. Documentationfrom a workshop held by the Federal Insti-tute of Labour (Bundesanstalt für Arbeit) inNuremberg on 26 and 27 March 1996. Bei-träge zur Arbeitsmarkt- und Berufsfor-schung 201. IAB, Nuremberg, pp. 11-45.

Glytsos N. P., 1990. Anticipated Graduate JobMismatch and Higher Education CapacityInadequacies in Greece. Higher Education,19(4), pp.397-418.

Grip A. de, 1989. The Economics of Overeducation:a Comment. Economics of Education Review,8(2), pp.205-207.

Groeneveld S., 1997. Passend meten. Overdefinities en metingen vab overscholing.Tijdschrift voor Arbeidsvraagstukken, 13(3),pp.273-282.

Groot W., 1993a. Overeducation and the Returnsto Enterprise-related Schooling. Economicsof Education Review, 12(4), pp.299-309.

Groot W., 1993b. Overscholing, onderscholing enhet rendement op bedrijfsopleiding. mens enmaatschappij, 68(4), pp.386-405.

Groot W., 1996. The incidence of, and returns toovereducation in the UK. Applied Econom-ics, 28, pp. 1345-1350.

Groot W., MaassenvandenBrink H., 1995. Alloca-tion and the Returns to Overeducation in theUnited Kingdom. Discussion Paper No95/205, Tinbergen Institute of the Universityof Amsterdam.

Groot W., Maassen van den Brink H., 1996.Overscholing en verdringing op de arbeids-markt. Economisch Statistische Berichte, 81,pp.74-77.

Groot W., Maassen van den Brink H., 1999.Overeducation in the labor market: measure-ment issues and effects (submitted manu-script).

Felix Büchel

534

Groot W., Maassen van den Brink H., forthcom-ing. Overeducation in the labor market: ameta-analysis. In: H. Oosterbeek, ed. Eco-nomics of Education Review, special issueon overeducation.

Gwartney J., 1970. Discrimination and IncomeDifferentials. American Economic Review,60,pp.396-408.

Halaby C.N., 1994. Overeducation and Skill Mis-match. Sociology of Education, 67(1), pp.47-59.

Handl J., 1996. Hat sich die berufliche Wertigkeitder Bildungsabschlüsse in den achtzigerJahren verringert? Eine Analyse der abhän-gig erwerbstätigen, deutschen Berufsanfän-ger auf der Basis von Mikrozensusergeb-nissen. Kölner Zeitschrift für Soziologie undSozialpsychologie, 48, pp.249-273.

Hartog J., 1980. Earnings and capability require-ments. Review of Economics and Statistics,62(2), pp.230-240.

Hartog J., 1981a. Personal Income Distribution. AMulticapability Theory. Nijhoff, Leiden, NL,and Boston, Mass., USA.

Hartog J., 1981b. Wages and allocation under im-perfect information. De Economist, 129(3),pp.311-323.

Hartog J., 1985a. Earnings Functions: Testing forthe Demand Side. Economics Letters, 19,pp.281-285.

Hartog J., 1985b. Overscholing? EconomischStatistische Berichten, 70, pp.152-156.

Hartog J., 1986. Allocation and the Earnings Func-tion. Empirical Economics, 11, pp.97-110.

Hartog J., 1997. Over-Education and Earnings.Where are we, where should we go? (submit-ted manuscript).

Hartog J., 1999a. Behind the veil of human capi-tal. OECD Observer, 215.

Hartog J., 1999b. On Returns to Education: Wan-dering Along the Hills of ORU Land. In: H.Heijke, J. Muysken, eds. Education andTraining in a Knowledge-based Economy.AEA-Macmillan Series, Macmillan Press,Basingstoke, UK, pp.3-45.

Hartog J., 1999c. Over- and under-education andthe relation to vocational training. EuropeanJournal for Vocational Training, 16(1),pp.47-52.

Hartog J., 1999d. Über- und Unterqualifikationund ihre Beziehung zur beruflichen Weiter-bildung. Europäische Zeitschrift für Berufs-bildung, 16(1), pp.52-58.

Hartog J., Jonker N., 1998. A Job to Match yourEducation: Does it Matter? In: H. Heijke, L.Borghans, eds. Towards a Transparent La-bour Market for Educational Decisions.Ashgate, Aldershot, pp.99-118.

Hartog J., Oosterbeek H., 1988. Education, Allo-cation and Earnings in the Netherlands:Overschooling? Economics of Education Re-view, 7(2), pp. 185-194.

Haugrund S., 1990. Qualifikationsreserven in derindustriellen Forschung und Entwicklung.Eine empirische Untersuchung zur Häufig-keit und zu ausgewählten Folgen einerarbeitsplatzbezogenen, individuellen Über-qualifikation. Poeschel, Stuttgart.

Hecker D.E., 1992. Reconciling conflicting data onjobs for college graduates. Monthly LaborReview, 115(7), pp.3-12.

Hecker D.E., 1995a. College graduates in ‘highschool’ jobs: a commentary. Monthly LaborReview, 115(12), pp.28ff.

Hecker D.E., 1995b. Further analyses of the labormarket for college graduates. Monthly LaborReview, 118(2), pp.39-41.

Heinelt H., Lohmann A., 1992. Immigranten imWohlfahrtsstaat am Beispiel der Rechts-positionen und Lebensverhältnisse von Aus-siedlern. Leske + Budrich, Opladen.

Henninges H. von, 1991. Bildungsstruktur, Aus-bildungsstruktur und ausbildungsadä-quater Einsatz. In: G. Wagner, B. vonRosenbladt, D. Blaschke, eds. An derSchwelle zur sozialen Marktwirtschaft. Er-gebnisse aus der Basiserhebung des sozio-ökonomischen Panels in der DDR im Juni1990. Beiträge zur Arbeitsmarkt- undBerufsforschung 143. Institute for Emp-loyment Research (IAB), Nuremberg, pp.17-22.

Henninges H. von, 1996. Steigende Qualifikations-anforderungen im Arbeiterbereich? Mittei-lungen aus der Arbeitsmarkt- und Berufs-forschung, 29(1), pp.73-82.

Hersch J., 1991. Education Match and Job Match.Review of Economics and Statistics, 79,pp.140-144.


535

Hersch J., 1995. Optimal ‘Mismatch’ and Promo-tions. Economic Inquiry, 33, pp.611-624.

Hofbauer H., Nagel E., 1987. Mobilität nach Ab-schluß der betrieblichen Berufsausbildung.Mitteilungen aus der Arbeitsmarkt- undBerufsforschung, 20(1), pp.45-73.

Hofmann C.F., 1995. EU-Arbeitsmarktumfrage:Aspekte der Arbeitsmärkte in Europa und inDeutschland. Federal Ministry of Labourand Social Affairs, Bonn.

Hofmann S., Vogeler G., 1995. Geschichtsstudiumund Beruf. Ergebnisse einer Befragung derAbsolventen des MagisterstudiengangesGeschichte in München 1987-1992. Students’Organization of Ludwig-Maximilians-Universität, Dept. of History, Munich.

House J., 1974. The effects of occupational stresson physical health. In: J. O’Toole, ed. Workand the Quality of Life. M.I.T. Press, Cam-bridge, Mass., USA, pp.145-170.

ILO, 1998. World Employment Report 1998-99:Employability in the global economy – howtraining matters. International Labour Of-fice, Geneva.

iwd, 1994a. Akademiker-Beschäftigung – EinFünftel unter Niveau. Informationsdienstdes Instituts der deutschen Wirtschaft (iwd),20(30), pp.4-5.

iwd, 1994b. Trainee-Programme: bei Akademikernbegehrt. Informationsdienst des Instituts derdeutschen Wirtschaft (iwd), 20(29), pp.4-5.

iwd, 1997. Einfach-Arbeitsplätze – Stiefkinder desArbeitsmarktes. Informationsdienst desInstituts der deutschen Wirtschaft (iwd),25(11), p. 6.

Jaffe A.J., Froomkin J., 1978. Occupational oppor-tunities for college-educated workers, 1950-75. Monthly Labor Review, 101, pp.15-21.

Jansen R., 1993. Ausbildung und Beschäftigung:Berufseinmündung und Wechselprozesse.In: R. Jansen, F. Stooss, eds. Qualifikationund Erwerbssituation im geeinten Deutsch-land. Ein Überblick über die Ergebnisse derBIBB/IAB-Erhebung 1991/92. Federal In-stitute for Vocational Training (BIBB), Ber-lin and Bonn, pp.24-43.

Jenkins D., 1974. Job Power. Heinemann, London.

Jensen U., 1999. Measuring overeducation withrobust stochastic earnings frontiers. Paper

presented at the EALE 1999 Conference,Regensburg, Germany, 23-26 September1999.

Johnson G. and W.R., 1992. Subjective underem-ployment and psychosocial stress: the roleof perceived social and supervisor support.The Journal of Social Psychology, 132, pp.11-22.

Johnson G. and W.R., 1996. Perceived Overquali-fication and Psychological Well-Being. TheJournal of Social Psychology, 136, pp. 435-445.

Johnson G.J. and W.R., 1997. Perceived Overquali-fication, Sense of Control and Depression.National Journal of Sociology, 11(1-2),pp.29-59.

Johnson G.J., Roy J.W., 1994. Perceived SkillUnderutilization and Distributive Justice:The Role of Perceived Supervisor Support.Sociological Spectrum, 14(2), pp.129-142.

Johnson G.J., Roy J.W., 1995. Subjective Under-employment and Job Satisfaction. Interna-tional Review of Modern Sociology, 25(1),pp.73-84.

Johnson W., 1978. A Theory of Job Shopping. Quar-terly Journal of Economics, 92, pp.261-277.

Jovanovic B., 1979a. Firm-specific Capital andTurnover. Journal of Political Economy,87(6), pp.1246-1260.

Jovanovic B., 1979b. Job Matching and the Theoryof Turnover. Journal of Political Economy,87(5), pp.972-990.

Kaiser M., Jelitto R., Otto M., Schwarz U., 1980.Berufliche Integration von Fachhochschul-absolventen. Mitteilungen aus der Arbeits-markt- und Berufsforschung, 13(1), pp.77-104.

Kalleberg A., Sørensen A.B., 1973. The Measure-ments of the Effects of Overtraining on JobAttitudes. Sociological Methods and Re-search, 2(2), pp.215-238.

Kasl S., 1974. Work and mental health. In:J.O’Toole, ed. Work and the Quality of Life.M.I.T.Press, Cambridge, Mass., USA,pp.171-196.

Kaufman R.T., 1984. A note on Richard Freeman’s‘Overinvestment in college training’. Jour-nal of Human Resources, 19(2), pp.278-279.

Felix Büchel

536

Keller B., 1997. Atypische Beschäftigungsverhält-nisse: Beschäftigungswirkungen und Bedin-gungen der Regulierung. In: D. Sadowski,K. Pull, eds. Vorschläge jenseits der Lohn-politik. Optionen für mehr Beschäftigung II.Campus, Frankfurt/Main and New York,pp.227-246.

Keller B., Klein T., 1994. Berufseinstieg und Mo-bilität von Akademikern zwischen Öffentli-chem Dienst und Privatwirtschaft. Evidenzaus der Konstanzer Absolventenbefragungbei Diplom-Verwaltungswissenschaftlern.Mitteilungen aus der Arbeitsmarkt- undBerufsforschung, 27(2), pp.152-160.

Khan L.J., Morrow P.C., 1991. Objective and Sub-jective Underemployment Relationships toJob Satisfaction. Journal of Business Re-search, 22, pp.211-218.

Kiker B.F., Santos M.C., deOliveira M.M., 1997.Overeducation and undereducation: evi-dence for Portugal. Economics of EducationReview, 16(2), pp.111-125.

Kornhauser A., 1965. Mental Health of the Indus-trial Worker. Wiley, New York.

Krais B., 1983. Thesen zur Entwicklung der Ar-beitsbedingungen von Akademikern in derprivaten Wirtschaft. In: Arbeitskreis Berufs-forschung, Bremen and Bergisch-Gladbach,Institut für Arbeitsmarkt- und Berufs-forschung, Nuremberg, Max-Planck-Institutfür Bildungsforschung, Berlin, and Wissen-schaftliches Zentrum für Berufs- und Hoch-schulforschung, Kassel, eds. Hochschul-expansion und Arbeitsmarkt. Problemstel-lungen und Forschungsperspektiven. Beiträ-ge zur Arbeitsmarkt- und Berufsforschung77. IAB, Nuremberg, pp. 34-47.

Krämer W., 1982. Der Zusammenhang zwischenBerufsausbildung und Berufstätigkeit. Eineempirische Untersuchung von Ausmaß, Ent-stehung und Folgen beruflicher Flexibilitätbei Führungskräften kleiner Unternehmen.Schwartz, Göttingen.

Kühlewind G., Tessaring M., 1975. Argumente fürund gegen eine beschäftigungsorientierteBildungspolitik. Verlag Otto Schwartz &Co., Göttingen.

Lambropoulos H., Psacharopoulos G., 1992. Edu-cational Expansion and Earnings Differen-tials in Greece. Comparative Education Re-view, 36(1), pp.52-70.

Levin H.M., 1995. Work and Education. In: M.Carnoy, ed. International Encyclopedia ofEconomics of Education. 2nd Edition.Pergamon Press, Oxford, UK, pp. 10-19.

Levin H.M., 1977. Book Review: The OvereducatedAmerican by Richard B. Freeman. HarvardEducational Review, 47, pp.226-231.

Lewin K., Cordier H., Heublein U., Dieter S., An-dermann H., 1994a. Studienabbruch,Hochschulwechsel und Examen im Winter-semester 1992/93 in Nordrhein-Westfalen:Hintergründe und Bewertung. Reportcommissioned by the Ministry of Science andResearch of the Federal State of NorthRhine/Westphalia, Düsseldorf.

Lewin K., Cordier H., Heublein U., Sommer D.,Andermann H., 1994b. Befragung derExmatrikulierten des Wintersemesters 1992/93 in Hessen. Vol. 7 of the Hochschulen se-ries, Ministry of Education and the Arts ofthe Federal State of Hesse, Wiesbaden.

Lichter D.T., 1988. Racial Differences in Under-employment in American Cities. AmericanJournal of Sociology, 93(4), pp.771-792.

Lippman S., McCall J., 1976a. The Economics ofJob Search. Part I. Optimal Search Policies.Economic Inquiry, 14 (June), pp.155-189.

Lippman S., McCall J., 1976b. The Economics ofJob Search. Part II. Empirical and PolicyImplications of Job Search. Economic In-quiry, 14 (September), pp.347-368.

Lucas R.E.B., 1977. Hedonic Wage Equations andPsychic Wages in the Returns to Schooling.American Economic Review, 67, pp.549-558.

Lutz B., Sengenberger W., 1974. Arbeitsmarkt-strukturen und öffentliche Arbeitsmarkt-politik. Eine kritische Analyse von Zielen undInstrumenten. Schwartz, Göttingen.

MacDonald G.M., 1982. A market equilibrumtheory of job assignment and sequential ac-cumulation of information. American Eco-nomic Review, 72(5), pp.1038-1055.

Madamba A.B., De Jong G.F., 1997. Job Mismatchamong Asians in the United States: EthnicGroup Comparisons. Social Science Quar-terly, 78(2), pp.524-542.

Mangione T.W., Quinn R.P., 1975. Job Satisfaction,Counterproductive Behavior, and Drug Useat Work. Journal of Applied Psychology, 60,pp.114-116.

Marsden D., 1997. Beschäftigungsförderungdurch arbeitsrechtliche Deregulierung? In:D. Sadowski, K. Pull, eds. Vorschläge jen-seits der Lohnpolitik. Optionen für mehr Be-


537

schäftigung II. Campus, Frankfurt/Mainand New York, pp.247-266.

Mason G., 1996. Graduate Utilisation in BritishIndustry: the Initial Impact of Mass HigherEducation. National Institute Economic Re-view, (May), pp.93-103.

McGoldrick K., Robst J., 1996. Gender Differencesin Overeducation. A Test of the Theory ofDifferential Overqualification. AmericanEconomic Review, 86, pp. 280-284.

Mendes de Oliveira M., Santos M.C., Kiker B.F.,(forthcoming). The role of human capital andtechnological change in overeducation. In:H.Oosterbeek, ed. Economics of EducationReview, special issue on overeducation.

Mertens D., ed., 1982. Konzepte der Arbeitsmarkt-und Berufsforschung. Eine Forschungs-inventur des IAB. Beiträge zur Arbeits-markt- und Berufsforschung 70. IAB,Nuremberg.

Mincer J., 1978. Family Migration Decisions. Jour-nal of Political Economy, 86(5), pp.749-773.

Mincer J., 1984. Overeducation or Underedu-cation? Comment. In: E. Dean, ed. Educa-tion and Economic Productivity. Ballinger,Cambridge, Mass., USA, pp.205-211.

Minks K.-H., 1992. Absolventenreport Wirtschafts-wissenschaften. Ergebnisse einer Untersu-chung zum Berufsübergang von Absolven-ten wirtschaftswissenschaftlicher Studien-gänge des Prüfungsjahres 1988/89. Vol.6/92 in the Bildung-Wissenschaft-Aktuellseries, Federal Ministry of Education andScience, Bonn.

Minks K.-H., 1996. Frauen aus technischen undnaturwissenschaftlichen Studiengängen.Ein Vergleich der Übergänge von Absolven-tinnen und Absolventen. Vol.116 in theHochschulplanung series, HIS, Hanover.

Minks K.-H., Filaretow B., 1993. Absolventen-report Sozialwissenschaften. Ergebnisse ei-ner Längsschnittuntersuchung zum Berufs-übergang von Absolventen sozialwissen-schaftlicher Diplom-Studiengänge. Vol.18/93 in the Bildung-Wissenschaft-Aktuellseries, Federal Ministry of Education andScience, Bonn.

Mortensen D.T., 1988. Wages, Separations, and JobTenure. On-the-Job Specific Training orMatching? Journal of Labor Economics, 6(4),pp.445-471.

Müller F., Beck M., 1993. Versteckte Arbeitslosig-keit als wirtschaftspolitisches Problem. De-finition und Messung am Beispiel bundes-deutscher Wasserversorgungsunternehmen.Jahrbuch für Sozialwissenschaft, 44(1),pp.53-77.

Muysken J., terWeel B., forthcoming. Overeduca-tion and Crowding out of Low-skilled Work-ers. In: L. Borghans, A. deGrip, eds. TheOvereducated Worker? The Economics ofSkill Utilization. Edward Elgar, Chelten-ham, UK, and Northampton, Mass., USA.

Nelson P., 1970. Information and ConsumerBehavior. Journal of Political Economy,78(2), pp.311-329.

Neubäumer R., 1993. Betriebliche Ausbildung‘über Bedarf ’ – empirische Ergebnisse undein humankapitaltheoretischer Ansatz.Jahrbuch für Sozialwissenschaft, 44,pp.104-131.

Neubäumer R., 1997. «Ausbildung über oder un-ter Bedarf» und die Finanzierung derDualen Ausbildung. Theoretische Erklä-rungen, empirische Ergebnisse und wirt-schaftspolitische Schlußfolgerungen. Paperpresented on 6 February 1997 to the an-nual meeting of the Committee on Educa-tional Economics (BildungsökonomischerAusschuss) of the Social Policy Association(Verein für Sozialpolitik). Submittedmanuscript, Potsdam.

Neubäumer R., forthcoming. Der Ausbildungs-stellenmarkt der Bundesrepublik Deutsch-land: Eine theoretische und empirische Ana-lyse. Volkswirtschaftliche Schriften, Duncker& Humblot, Berlin.

Neusel A., 1983. Praxisbezug des Studiums alsArbeitsmarktstrategie für Hochschul-absolventen? In: Arbeitskreis Berufs-forschung, Bremen and Bergisch-Gladbach,Institut für Arbeitsmarkt- und Berufs-forschung, Nuremberg, Max-Planck-Institutfür Bildungsforschung, Berlin, and Wissen-schaft-liches Zentrum für Berufs- undHochschulforschung, Kassel, eds. Hoch-schulexpansion und Arbeitsmarkt. Problem-stellungen und Forschungsperspektiven.Beiträge zur Arbeitsmarkt- und Berufs-forschung 77. IAB, Nuremberg, pp.235-258.

Nuthmann R., 1983. Verbleib von Hochschulabsol-venten, Annahmen zur Entwicklung vonTätigkeiten und Arbeitsbedingungen undArbeitsformen zwischen ‘richtiger’ Beschäf-tigung und Arbeitslosigkeit. In: ArbeitskreisBerufsforschung, Bremen and Bergisch-

Felix Büchel

538

Gladbach, Institut für Arbeitsmarkt- undBerufsforschung, Nuremberg, Max-Planck-Institut für Bildungsforschung, Berlin, andWissenschaftliches Zentrum für Berufs- undHochschulforschung, Kassel, eds. Hoch-schulexpansion und Arbeitsmarkt. Problem-stellungen und Forschungsperspektiven.Beiträge zur Arbeitsmarkt- und Berufs-forschung 77. IAB, Nuremberg, pp.1-19.

Oaxaca R., 1973. Male-Female Wage Differentialsin Urban Labor Markets. International Eco-nomic Review, 14, pp.693-709.

OECD, 1995a. Employment Outlook, July 1995.Organization for Economic Cooperation andDevelopment, Paris.

OECD, 1995b. Making Education Count. Develop-ing and Using International Indicators. Or-ganization for Economic Cooperation andDevelopment, Paris.

Ofek H., Merrill Y., 1997. Labor immobility andthe formation of gender wage gaps in localmarkets. Economic Inquiry, 35(1), pp.28-47.

Ofner F., 1994. Karriere durch Lehre? ZurBerufseinmündung von Absolventen derLehrausbildung. Wirtschaft und Gesell-schaft, 20(3), pp.337-360.

Omori Y., 1997. Stigma Effects of Nonemployment.Economic Inquiry, 35, pp.394-416.

Oosterbeek H., 1986. Overscholing en inkomen. In:J. Hartog, J. Ritsen, eds. Economischeaspecten van onderwijs. Swets & Zeitlinger,Lisse, NL, pp.120-128.

Oosterbeek H., Webbink D., 1996. Over scholing,overscholing en inkomen. EconomischStatistische Berichten, 81, pp.240-241.

Ours J.C. van, Ridder G., 1995. Job matching andjob competition: are lower-educated work-ers at the back of job queues? European Eco-nomic Review, 39, pp.1717-1731.

Patrinos H.A., 1995. Socioeconomic Background,Schooling, Experience, Ability and MonetaryRewards in Greece. Economics of EducationReview, 14 (1), pp.85-91.

Patrinos, H.A., 1997. Overeducation in Greece.International Review of Education, 43(2-3),1997, pp.203-223.

Pfeiffer F., Blechinger D., 1995. Verwertbarkeit derBerufsausbildung im technischen Wandel.In: W. Franz, V. Steiner, eds. Der west-deutsche Arbeitsmarkt im strukturellenAnpassungsprozeß. Centre for European

Economic Research (ZEW), Economic Analy-sis No 3, pp. 105-140.

Plicht H., Schober K., Schreyer F., 1994. ZurAusbildungsadäquanz der Beschäftigungvon Hochschulabsolventinnen (und -absol-venten). Versuch einer Quantifizierunganhand der Mikrozensen 1985 bis 1991.Mitteilungen aus der Arbeitsmarkt- undBerufsforschung, 27(3), pp.177-204.

ProjektgruppePanel, Das Sozio-ökonomische Panel(SOEP) im Jahre 1994. Vierteljahrshefte zurWirtschaftsforschung, 1, 1995, pp.5-15.

Psacharopoulos G., 1988. Earnings and Educationin Greece, 1960-1977. European EconomicReview, 17, pp.333-347.

Quinn R., Mandilovitch M., 1995. Education andJob Satisfaction: a Questionable Payoff. Sur-vey Research Center, University of Michi-gan, AnnArbor, Mich., USA.

Rawlins V.L., Ulman L., 1974. The utilization ofcollege-trained manpower in the UnitedStates. In: Gordon M. S., ed. Higher Educa-tion and the Labor Market. New York:MacGraw-Hill, pp. 195-235.

Rippe W., 1984. Die Analyse von betrieblichenEntscheidungsprozessen bei der Nachfragevon Hochschulabsolventen. Ein zu Rechtvernachlässigter Ansatz bei der Erfor-schung von Absorptionsmechanismen desArbeitsmarkts? In: H.P. Widmaier, ed. DasArbeitskräfteangebot zwischen Markt undPlan. Duncker & Humblot, Berlin, pp.61-96.

Rippe W., 1988. Zur Ausbildungsadäquanz der Ar-beitsplätze von Hochschulabsolventen. Kon-zeptionen und Indikatoren, Erfahrungenund Erwartungen’, In: H.-J. Bodenhöfer, ed.Bildung, Beruf, Arbeitsmarkt. Documents ofthe Social Policy Association (Verein für So-cialpolitik), N.F. 174. Duncker & Humblot,Berlin, pp.171-208.

Robst J., 1995a. Career mobility, job match, andovereducation. Eastern Economic Journal,21, pp.539-550.

Robst J., 1995b. College Quality and Overedu-cation. Economics of Education Review, 14,pp.221-228.

Rosen S., 1972. Learning and Experience in theLabor Market. Journal of Human Resources,7(3), pp.326-342.

Rosen S., 1978. Substitution and Division of Labor.Economica, 45, pp.235-250.


539

Rosenfeld R.A., 1992. Job mobility and career proc-esses. Annual Review of Sociology, 18, pp.39-61.

Roy A.D., 1951. Some Thoughts on the Distribu-tion of Earnings. Oxford Economic Papers,3, pp.135-146.

Ruiz-Quintanilla S.A., Claes R., 1996. Determi-nants of underemployment of young adults:a multi-country study. Industrial and laborrelations review, 49(3), pp.424-438.

Rumberger R.W., 1980. The Economic Decline ofCollege Graduates: Fact or Fallacy? Jour-nal of Human Resources, 15, pp.99-122.

Rumberger R.W., 1981a. The Changing Skill Re-quirements of Jobs in the U.S. Economy.Industrial and Labor Relations Review,34(4), pp.578-590.

Rumberger R.W., 1981a. Overeducation in the U.S.Labor Market. Praeger, New York.

Rumberger R.W., 1981c. The Rising Incidence ofOvereducation in the U.S. Labor Market.Economics of Education Review, 1(3),pp.293-314.

Rumberger R.W., 1984. The changing economicbenefits of college graduates. Economics ofEducation Review, 3, pp.3-11.

Rumberger R.W., 1987. The Impact of SurplusSchooling on Productivity and Earnings.Journal of Human Resources, 22(1), pp. 24-50.

Rumberger R.W., 1994. Labour market outcomesas indicators of educational performance. In:OECD, Centre for Educational Research andInnovation, ed. Making Education Count.Developing and Using International Indica-tors. OECD, Paris, pp.265-286.

Sattinger M., 1975. Comparative advantage andthe distribution of earnings and abilities.Econometrica, 43(3), pp.455-468.

Sattinger M., 1977. Compensating Wage Differ-ences. Journal of Economic Theory, 16,pp.496-503.

Sattinger M., 1980. Capital and the distribution oflabor earnings. North-Holland, Amsterdam.

Schasse U., 1991. Betriebszugehörigkeitsdauer undMobilität. Eine empirische Untersuchung zurStabilität von Beschäftigungsverhältnissen.Campus, Frankfurt/Main and New York.

Schlaffke W., 1972. Akademisches Proletariat?Fromm, Osnabrück, Germany.

Schlegelmilch C., 1982. Grauer Arbeitsmarkt fürHochschulabsolventen. Zur Typologie vonGrauzonenbeschäftigungen und Problemenihrer empirischen Erfassung. Soziale Welt,33(3/4), pp.400-430.

Schlegelmilch C., 1983a. Arbeitsorientierungenund Lebensperspektiven von Hochschul-absolventen in der Grauzone. In: J. Matthes,ed. Krise der Arbeitsgesellschaft? Campus,Frankfurt/Main and New York.

Schlegelmilch C., 1983b. ‘Grauzonen’ des sozialenVerbleibs von Hochschulabsolventen alsForschungsproblem. In: Arbeitskreis Berufs-forschung, Bremen and Bergisch-Gladbach,Institut für Arbeitsmarkt- und Berufs-forschung, Nuremberg, Max-Planck-Institutfür Bildungsforschung, Berlin, and Wissen-schaftliches Zentrum für Berufs- undHochschulforschung, Kassel, eds. Hoch-schulexpansion und Arbeitsmarkt. Problem-stellungen und Forschungsperspektiven.Beiträge zur Arbeitsmarkt- und Berufs-forschung 77. IAB, Nuremberg, pp. 48-59.

Schlegelmilch C., 1987. Taxifahrer Dr. phil.Akademiker in der Grauzone des Arbeits-marktes. Leske + Budrich, Opladen, Ger-many.

Schreyer F., 1999. Der Akademikerarbeitsmarkt– ein Überblick. Informationen für dieBeratungs- und Vermittlungsdienste derBundesanstalt für Arbeit, 28/99, pp.2275-2283.

Schwartz E., Thornton R., 1980. Overinvestmentin College Training? Journal of Human Re-sources, 15(1), pp.121-123.

Schwarze J., 1993. Qualifikation, Überqualifi-kation und Phasen des Transformations-prozesses. Die Entwicklung der Lohn-struktur in den neuen Bundesländern. Jahr-bücher für Nationalökonomie und Statistik,211(1-2), pp.90-107.

Scoville J.G., 1966. Education and Training Re-quirements for Occupations. Review of Eco-nomics and Statistics, 48, pp.387-394.

Shelley K.J., 1992. The future of jobs for collegegraduates. Monthly Labor Review, 115(7),pp.13-21.

Shelley K.J., 1994. More Job Openings – EvenMore New Entrants: the Outlook for CollegeGraduates, 1992 – 2005. Occupational Out-look Quarterly, 20 (summer), pp.4-9.

Sheppard H.L., Herrick N.Q., 1972. Where HaveAll the Robots Gone: Worker Dissatisfactionin the 70s. Free Press, New York.

Felix Büchel

540

Shockey J.W., 1989. Overeducation and Earnings:A Structural Approach to Differential At-tainment in the U.S. Labor Force (1970-1982). American Sociological Review, 54(5),pp.856-864.

Sicherman N., 1987. Occupational Mobility andHuman Capital Theory. Ph.D. Dissertation,Dept. of Economics, Columbia University,New York.

Sicherman N., 1991. ‘Overeducation’ in the LaborMarket. Journal of Labor Economics, 9(2),pp.101-122.

Sicherman N., Galor O., 1990. A Theory of CareerMobility. Journal of Political Economy,98(1), pp.169-192.

Sloane P., Battu H., Seaman P.T., forthcoming.Over-education, Undereducation and theBritish Labour Market. Applied Econom-ics.

Sloane P.J., Battu H., Seaman P.T., 1996. Overedu-cation and the Formal Education/Experienceand Training Trade-Off. Applied EconomicsLetters, 3(8), pp.511-515.

Smith H.L., 1986. Overeducation and Underem-ployment: An Agnostic Review. Sociology ofEducation, 59, pp.85-99.

Smith J.P., Welch F., 1978. The overeducatedAmerican? A review article. RAND Discus-sion Paper No147.

Sørensen A.B., 1977. The Structure of Inequalityand the Process of Attainment. AmericanSociological Review, 42 (December), pp.965-978.

Spence M., 1973. Job Market Signaling. The Quar-terly Journal of Economics, 88, pp.355-374.

Spence M., 1974. Market Signalling: InformationTransfer in Hiring and Related ScreeningProcesses. Harvard University Press, Cam-bridge, Mass., USA.

Spiegel: see Der Spiegel.

Spilerman S., Lunde T., 1991. Features of Educa-tional Attainment and Job Promotion Pros-pects. American Journal of Sociology, 97(3),pp.689-720.

Staines G.L., Quinn R.P., 1979. American workersevaluate the quality of their jobs. MonthlyLabor Review, 102(3-12).

Stifterverband für die Deutsche Wissenschaft, ed.,1993. Qualifikationsanforderungen anHochschulabsolventen – Empfehlungen ausSicht der Wirtschaft. Essen, (brochure).

Stigler G.J., 1961. The Economics of Information.Journal of Political Economy, 3, pp.213-225.

Stiglitz J.E., 1975. The Theory of ‘Screening’, Edu-cation, and the Distribution of Income. TheAmerican Economic Review, 65(3), pp.283-300.

Stooss F., 1979. Ausgewählte Befunde zur Situati-on der Hochschulabsolventen in der Bundes-republik Deutschland im Jahre 1979. Zu-sammenfassung empirischer Ergebnisse auseiner Repräsentativ-Befragung der InfratestWirtschaftsforschung GmbH, München.Mitteilungen aus der Arbeitsmarkt- undBerufsforschung, 12(4), pp.607-624.

Suda Z., 1979. Universal Growth of EducationalAspirations and the ‘Overqualification’ Prob-lem. European Journal of Education, 14(2),pp.113-164.

Sullivan T.A., 1978. Marginal Workers, MarginalJobs: The Underutilization of AmericanWorkers. University of Texas Press, Austin,Texas, USA.

Sullivan T.A., 1981. Sociological Views of LaborMarkets: Some Missed Opportunities andNeglected Directions. In: I. Berg, ed. Socio-logical Perspectives on Labor Markets. Aca-demic Press, New York, etc., pp.329-346.

Szydlik M., 1996a. Ethnische Ungleichheit auf demdeutschen Arbeitsmarkt. Kölner Zeitschriftfür Soziologie und Sozialpsychologie, 48(4),pp. 658-676.

Szydlik M., 1996b. Zur Übereinstimmung vonAusbildung und Arbeitsplatzanforderung inder Bundesrepublik Deutschland. Mittei-lungen aus der Arbeitsmarkt- und Berufs-forschung, 29(2), pp.295-306.

Szydlik M., 1997a. Ausbildung und Beschäftigungvon Ost- und Westdeutschen. Aus Politikund Zeitgeschichte, supplement to the 13June 1997 edition of the weekly newspaperDas Parlament, pp.13-24.

Szydlik M., 1997b. Consequences of AllocationProcesses in the Labor Market: a compari-son of the United States and Germany. In:T.A. Dunn, J. Schwarze, eds. Proceedings ofthe 1996 Second International Conference ofthe German Socio-Economic Panels StudyUsers, special issue of the Vierteljahrsheftezur Wirtschaftsforschung, Vol.66, No 1,pp.41-46.


541

Tessaring M., 1978. Zum Verbleib der Hochschul-absolventen seit 1971. Mitteilungen aus derArbeitsmarkt- und Berufsforschung, 11(2),pp.171-177.

Tessaring M., 1980. Evaluation von Bildungs- undQualifikationsprognosen, insbesondere fürhochqualifizierte Arbeitskräfte. Mittei-lungen aus der Arbeitsmarkt- und Berufs-forschung, 13(3), pp.374-397.

Tessaring M., 1981. Arbeitslosigkeit, Verbleib undBeschäftigungsmöglichkeiten der Hoch-schulabsolventen. Mitteilungen aus derArbeitsmarkt- und Berufsforschung, 141(4),pp.391-404.

Tessaring M., 1982. Die Bilanzierung von Angebotund Bedarf an hochqualifizierten Arbeits-kräften. In: D. Mertens, ed. Konzepte derArbeitsmarkt- und Berufsforschung. EineForschungsinventur des IAB. Beiträge zurArbeitsmarkt- und Berufsforschung 70. IAB,Nuremberg.

Tessaring M., 1983. Verbleib und Beschäftigungs-potentiale von Hochschulabsolventen – The-sen. In: Arbeitskreis Berufsforschung, Bre-men and Bergisch-Gladbach, Institut fürArbeitsmarkt- und Berufsforschung, Nu-remberg, Max-Planck-Institut für Bildungs-forschung, Berlin, and WissenschaftlichesZentrum für Berufs- und Hochschulfor-schung, Kassel, eds. Hochschulexpansionund Arbeitsmarkt. Problemstellungen undForschungsperspektiven. Beiträge zur Ar-beitsmarkt- und Berufsforschung 77. IAB,Nuremberg, pp.20-33.

Tessaring M., 1984. Akademikerbeschäftigung imWandel. In: H.P. Widmaier, ed. Das Arbeits-kräfteangebot zwischen Markt und Plan.Duncker & Humblot, Berlin, pp.251-280.

Tessaring M., 1994a. Indikatoren der Akademiker-beschäftigung. In: C. Konegen-Grenier, W.Schlaffke, eds. Akademikerbeschäftigungund Akademikerbedarf. Deutscher Instituts-Verlag, Cologne, pp.38-54.

Tessaring M., 1994b. Langfristige Tendenzen desArbeitskräftebedarfs nach Tätigkeiten undQualifikationen in den alten Bundesländernbis zum Jahre 2010. Eine erste Aktu-alisierung der IAB/Prognos-Projektionen1989/91. Mitteilungen aus der Arbeitsmarkt-und Berufsforschung, 27(1), pp.5-19.

Szydlik M., 1997c. Skill Mismatch in the UnitedStates of America, the former Federal Repub-lic of Germany and the German DemocraticRepublic. Research Report 59 of the Re-search Group on Aging and the Life Course(Forschungsgruppe Altern und Lebenslauf(FALL)), Free University of Berlin.

Szydlik M., 1998. Vocational Training and the Qual-ity of Immigrant and Minority Jobs in Ger-many and the United States. In: H. Kurthen,J. Fijalkowski, G.G. Wagner, eds. Immigra-tion, Citizenship, and the Welfare State inGermany and the United States: ImmigrantIncorporation. PartA, Vol.14(A) of the Indus-trial Development and Social Fabric Series.JAI Press, Stanford, Conn., USA, pp.119-135.

Taubman P., Wales T., 1974. Higher Education andEarnings: College as an investment and ascreening device. McGraw-Hill, New York.

Teichler U., 1988. Hochschule und Beruf in Euro-pa. In: L. Reyher, J. Kühl, eds. Resonanzen.Arbeitsmarkt und Beruf – Forschung undPolitik. Beiträge zur Arbeitsmarkt- undBerufsforschung 111. Institute for Employ-ment Research (IAB), Nuremberg, pp.450-495.

Teichler U., 1994. Zur Akademikerbeschäftigungund zum Akademikerbedarf im internatio-nalen Vergleich. In: C. Konegen-Grenier, W.Schlaffke, eds. Akademikerbeschäftigungund Akademikerbedarf. Deutscher Instituts-Verlag, Cologne, pp.21-37.

Teichler U., 1996. Higher Education and Gradu-ate Employment in Europe. Select Findingsfrom Previous Decades. Workshop Report No52 of the Academic Centre for Occupationaland Higher Education of Kassel University/Polytechnic, Kassel, Germany.

Teichler U., 1998. Der Wandel der Beschäftigungund Berufstätigkeit von Hochschulabsol-venten in Deutschland. In: A. Takanashi, U.Teichler, eds. Berufliche Kompetenz-entwicklung im Bildungs- und Beschäf-tigungssystem in Japan und Deutschland.Frankfurter Studien zur Bildungsforschung,Nomos, Baden-Baden, pp.83-111.

Teichler U., Schomburg H., Winkler H., 1992. Stu-dium und Berufsweg von Hochschulabsol-venten. Vol.18/92 in the Bildung – Wissen-schaft – Aktuell series, Federal Ministry ofEducation and Science, Bonn.

Terry S.L., 1981. Involuntary part-time work: newinformation from the CPS. Monthly LaborReview, 104(2), pp.70-74.

Felix Büchel

542

Tessaring M., 1998. The Future of Work and Skills– Visions, Trends and Forecasts. In: Cedefop,ed.Vocational education and training – theEuropean research field. Background report1998. Office for Official Publications of theEuropean Communities, Luxembourg,pp.271-318.

Thurow L.C., 1975. Generating Inequality. Mecha-nisms of Distribution in the U.S. Economy.Basic Books, New York.

Tinbergen J., 1956. On the Theory of Income Dis-tribution. Weltwirtschaftliches Archiv, 77,pp.155-175.

Topel R.H., 1986. Local Labor Markets. Journalof Political Economy, 94(3), pp.111-143.

Topel R.H., Ward M.P., 1992. Job mobility and thecareers of young men. The Quarterly Jour-nal of Economics, 107(2), pp.439-479.

Tsang M.C., 1987. The Impact of Underutilizationof Education on Productivity: A Case Studyof the U.S. Bell Companies. Economics ofEducation Review, 6(3), pp.239-254.

Tsang M.C., Levin H.M., 1985. The Economics ofOvereducation. Economics of Education Re-view, 4(2), pp.3-104.

Tsang M.C., Levin H.M., 1989. The Economics ofOvereducation: Reply. Economics of Educa-tion Review, 8(2), pp. 209 ff.

Tsang M.C., Rumberger W.R., Levin H.M., 1991.The Impact of Surplus Schooling on WorkerProductivity. Industrial Relations, 30(2),pp.209-228.

Tsoucalas C., 1981. Some Aspects of ‘Overedu-cation’ in Modern Greece. Journal of theHellenic Diaspora, 8(1-2), pp.109-122.

Tyler J., Murnane R.J., Levy F.S., 1995a. Are lotsof college graduates taking high school jobs?A reconsideration of the evidence. NBERWorking Papers No 5127.

Tyler J., Murnane R.J., Levy F.S., 1995b. Are morecollege graduates really taking high schooljobs? Monthly Labor Review, 118(12), pp.18-27.

Tyler J., Murnane R.J., Levy F.S., 1995c. Did thejob market worsen for male college gradu-ates over the 1980s? Different answers for

different age groups. Research in Labor Eco-nomics, 14, pp.69-89.

Vahey S.P., forthcoming. The Great CanadianTraining Robbery: Evidence on the Returnsto Educational Mismatch. In: H. Oosterbeek,ed. Economics of Education Review, specialissue on overeducation.

Velden R.K.W van der, Smoorenburg M.S.M., 1999.The Measurement of Overeducation and Un-der-education: Self-Report vs. Job-AnalystMethod. Paper submitted to the LasmasSéminaire ‘Catégarisations des statutsd´emploi ed d´activité. Paris, France. 15 No-vember 1999.

Velling J., Pfeiffer F., 1997. Arbeitslosigkeit,inadäquate Beschäftigung, Berufswechselund Erwerbsbeteiligung. Centre for Euro-pean Economic Research (ZEW) Documen-tation No 97-02, Mannheim, Germany.

Verdugo R.R., Turner Verdugo N., 1988. Over-education and the Earnings of Black, His-panic, and White Male Workers. Sociologi-cal Perspectives, 3(2), pp.190-212.

Verdugo R.R., Turner Verdugo N., 1989. The Im-pact of Surplus Schooling on Earnings: SomeAdditional Findings. The Journal of HumanResources, 24(4), pp.629-643.

Verdugo R.R., Turner Verdugo N., 1992. SurplusSchooling and Earnings: Reply to Cohn andto Gill and Solberg. The Journal of HumanResources, 27(4), pp.690-695.

Vincens J., 1995. Graduates and the Labour Mar-ket in France. European Journal of Educa-tion, 30(2), pp.133-156.

Von Henninges: see Henninges.

Vroom V., 1964. Work and Motivation. Wiley, NewYork.

Weißhuhn G., 1978. Beschäftigungschancen undQualifikation. Zur Stabilität des Arbeits-marktes bei Bildungsexpansion und Wandelder Arbeitsplatzanforderungen. Campus,Frankfurt/Main.

Weißhuhn G., 1996. Szenarien der Akade-mikerbeschäftigung im Rahmen der IAB-Projektionen des Qualifikationsbedarfs. In:M. Tessaring, ed. Die Zukunft der Akade-mikerbeschäftigung. Documentation from aworkshop held by the Federal Institute ofLabour (Bundesanstalt für Arbeit) inNuremberg on 26 and 27 March 1996. Bei-


543

träge zur Arbeitsmarkt- und Berufsfor-schung 201. IAB, Nuremberg, pp.69-103.

Weißhuhn G., Büchel F., 1998. Germany. In:OECD, ed. Pathways and Participation inVocational and Technical Education andTraining, commissioned by the Federal Min-istry of Education and Science, Bonn. OECD,Paris, pp. 195-239.

Weißhuhn G., Wahse J., König A., 1994. Arbeits-kräftebedarf in Deutschland bis 2010.Arbeitskräfteeinsatz 1978 – 1990 und Sze-narien bis 2010 für die alten und neuen Bun-desländer. Vol.116 in the Studien zu Bildungund Wissenschaft series. Bock, Bonn.

Wissenschaftsrat, 1999. Stellungnahme zum Ver-hältnis von Hochschulausbildung undBeschäftigungssystem. Opinion delivered on9 July 1999. Public Relations Division,Scientific Council (Wissenschaftsrat), Würz-burg, Germany.

Witmer D.R., 1980. Has the Golden Age of Ameri-can Higher Education Come to an AbruptEnd? Journal of Human Resources, 15(1),pp.113-120.

Witte J.C., 1999. Education and Training in aChanging Labor Market. Youth Employmentin the United States and Germany: TheMechanics of Restructuring. In: T.A. Dunn,J.R. Frick, J.C. Witte, eds. Proceedings ofthe 1998 Third International Conference ofthe German Socio-Economic Panels StudyUsers. Special issue of Vierteljahrshefte zurWirtschaftsforschung. Duncker& Humblot,Berlin, pp.191-198.

Witte J.C., Kalleberg A.L., 1995. Matching Train-ing and Jobs: the Fit Between VocationalEducation and Employment in the GermanLabor Market. European Sociological Re-view, 11(3), pp.293-317.

Witte M. de, Steijn A.J., 1998. Automation, Un-deremployment and Workers’ Satisfaction.Paper submitted to the WES. Cambridge.

Wonneberger B., 1994. Fehlende Lehrstellen undFehlqualifizierung als Problem des dualenSystems sowie deren berufsbildungspoli-tische Behandlung durch Staat, Arbeitgeberund Gewerkschaften in den Jahren 1978 –1986: strukturelle Mängel der Facharbeiter-ausbildung. European University Studies,SeriesV – economics and business ma-nagement, Lang, Frankfurt/Main.

Zemsky R., 1998. Labor, Markets, and EducationalRestructuring. In: R. Zemsky, P. Cappelli,eds. The Changing Educational Quality ofThe Workforce. Special issue, Annals of theAmerican Academy of Political and SocialSciences (AAPSS), Vol. 559, pp.77-90.

Zhou M., 1993. Underemployment and EconomicDisparities among Minority Groups. Popu-lation Research & Policy Review, 12(2),pp.139-157.

Zwick T., 2000. Market Power of Skilled WorkersCauses Unemployment – So What? In: P. deGijsel, R. Olthoff, T. Zwick, eds. The Unem-ployment Debate: Current Issues. Metropo-lis, Marburg, pp. 105-124.

Zwick T., forthcoming. Over-qualification MakesLow-wage Employment Attractive. In: L.Borghans, A. deGrip, eds. The OvereducatedWorker? The Economics of Skill Utilization.Edward Elgar, Cheltenham, UK, and North-ampton, Mass., USA.

Felix Büchel

544

ad: Adequate employment = congruence between job and education (includingunderqualification).A12ov: Overqualification = underemployment.+: Degree of mismatch not clearly determinable.– : Implausible combination.

1) including: Apprenticeship degree (Lehre); in former GDR: includingBerufsausbildung/Facharbeiterabschluß.Master craftsmen degree [former GDR] (Meister DDR)Full-time vocational school degree (Berufsfachschule)School of public health degree (Schule desGesundheitswesens)Technical college degree [FRG] (Fachschule)School for civil servants (Beamtenausbildung)Other vocational degree (sonstiger berufsbildender Abschluß)

2) including: Technical college degree [former GDR] (Ingenieur-/Fachschule DDR)Post-sec. technical college degree (Fachhochschule)University degree (Hochschule/Universität)

Other qualification levels used in Tables 1 to 14:Lower general schooling (Hauptschulabschluß / 8. Klasse)Intermediate general schooling (Realschulabschluß / 10. Klasse; includingFachabitur)Higher general schooling (Abitur)

For East German Classification see Büchel/Weißhuhn (1997a).

Source: Büchel/Weißhuhn (1997a).

Unskilled/semi-skilled worker ov ovSkilled worker/foreman/master craftsman + –White-collar worker in an unskilled job ov ovWhite-collar worker in a skilled job + +White-collar worker in a highly skilled job – –Self-employed person ov ovCivil servant – –Unskilled/semi-skilled worker ov ovSkilled worker/foreman/master craftsman + –White-collar worker in an unskilled job ov ovWhite-collar worker in a skilled job + +White-collar worker in a highly skilled job + +Self-employed person ov ovCivil servant + –Unskilled/semi-skilled worker + ovSkilled worker/foreman/master craftsman ad ovWhite-collar worker in an unskilled job ad ovWhite-collar worker in a skilled job ad ovWhite-collar worker in a highly skilled job ad adSelf-employed person ad ovCivil servant ad adUnskilled/semi-skilled worker – –Skilled worker/foreman/master craftsman – –White-collar worker in an unskilled job – –White-collar worker in a skilled job – +White-collar worker in a highly skilled job ad adSelf-employed person ad adCivil servant – ad

No specialtraining required/Only a shortinductionon the job

A lengthy periodof coachingat my placeof work

Attendance atspecial theoreticalor practicalcourses/Acertificate ofvocational training

A university orpost-sec. technicalcollege degree

Classification model of overqualification (W est Germany , qualification level by job-requirement levelby occupational status)

Job-requirement Occupational status Classification with regard to thelevel degree of congruence between job

and educationQualification level gained

Vocational University ordegree 1) post-secondary

technical collegedegree 2)

Annex


545

Tabl

e 1:

Wor

kers

bro

ken

dow

n by

form

al v

ocat

iona

l ski

lls, t

he d

egre

e of

con

grue

nce

betw

een

job

and

educ

atio

n, a

nd v

ario

us jo

bch

arac

teris

tics

and

grou

p-re

late

d av

erag

es fr

om th

ese

char

acte

ristic

gro

ups

(Wes

t Ger

man

y an

d E

ast G

erm

any

, 199

5).

Job

char

acte

ristic

sW

est G

erm

any

Eas

t Ger

man

y

(With

out

(With

out

any

vo-

any

vo-

catio

nal

over

-ad

e-ov

er-

ade-

catio

nal

over

-ad

e-ov

er-

ade-

degr

ee)

qual

ified

quat

equ

alifi

edqu

ate

degr

ee)

qual

ified

quat

equ

alifi

edqu

ate

Pay

cha

ract

eris

tics:

– po

tent

ial g

ross

mon

thly

sal

ary

(ave

rage

, in

DM

)13,

397

3,23

54,

499

4,80

67,

162

2,41

22,

413

2,95

83,

101

4,50

8–

actu

al n

et m

onth

ly s

alar

y (a

vera

ge, i

n D

M)

1,95

61,

917

2,76

12,

084

4,60

51,

419

1,50

92,

023

2,02

62,

894

– ne

t hou

rly p

ay (

aver

age

in D

M)2

13.0

12.3

16.8

14.8

27.5

9.3

9.6

11.2

11.1

15.3

– pa

y so

cial

sec

urity

con

trib

utio

ns3

84%

84%

91%

72%

74%

84%

96%

98%

99%

96%

– no

agr

eed

wag

e st

ruct

ure;

free

ly a

gree

d w

ages

27%

31%

21%

69%

18%

32%

30%

29%

39%

20%

– pa

y de

term

ined

by

leng

th o

f ser

vice

436

%26

%49

%26

%66

%28

%27

%42

%47

%61

%–

have

ent

itlem

ent t

o co

mpa

ny p

ensi

on5

26%

20%

32%

21%

30%

4%2%

7%6%

11%

Occ

upat

iona

l sta

tus:

– B

lue-

colla

r w

orke

r60

%55

%23

%19

%0%

61%

73%

49%

19%

0%–

Whi

te-c

olla

r w

orke

r30

%25

%59

%47

%53

%34

%21

%41

%65

%81

%–

Civ

il se

rvan

t2%

0%7%

0%36

%2%

0%2%

0%8%

– S

elf-

empl

oyed

69%

20%

10%

34%

11%

2%6%

8%17

%12

%E

mpl

oym

ent s

tatu

s:–

full-

time

empl

oym

ent

71%

70%

84%

61%

89%

70%

79%

92%

80%

89%

– pa

rt-t

ime

empl

oym

ent

18%

21%

14%

18%

9%18

%17

%7%

19%

9%–

min

imal

/irre

gula

r em

ploy

men

t11

%9%

2%21

%2%

12%

4%0%

1%1%

Wor

king

tim

e:–

agre

ed w

eekl

y w

orki

ng ti

me

(ave

rage

, in

hour

s)33

.133

.435

.536

.434

.934

.136

.439

.136

.537

.5–

no fi

rmly

agr

eed

wor

king

tim

e15

%26

%13

%44

%21

%20

%10

%10

%20

%14

%–

actu

al w

eekl

y w

orki

ng ti

me

(ave

rage

, in

hour

s)35

.336

.239

.230

.241

.937

.439

.743

.941

.545

.2–

over

time

wor

ked

durin

g th

e la

st m

onth

(av

erag

e, in

hou

rs)

5.3

6.8

9.7

4.0

15.6

5.5

6.6

8.4

12.1

15.4

– nu

mbe

r of

agr

eed

wor

king

day

s pe

r w

eek

(ave

rage

, in

days

)4.

95.

15.

04.

95.

15.

05.

15.

15.

15.

1–

varia

ble/

no fi

xed

num

ber

of w

orki

ng d

ays

per

wee

k9%

12%

10%

33%

9%7%

5%10

%12

%7%

– nu

mbe

r of

wor

king

hou

rs p

er d

ay, i

f fix

ed (

aver

age,

in h

ours

)7.

37.

37.

98.

08.

67.

87.

78.

58.

48.

8–

chan

geab

le/n

o fix

ed n

umbe

r of

hou

rs p

er d

ay8%

16%

10%

26%

11%

11%

7%9%

10%

9%–

Wor

k in

the

even

ing

or a

t nig

ht (

afte

r 19

.00)

17%

24%

22%

13%

19%

24%

20%

27%

18%

22%

– W

eeke

nd w

orki

ng (

Sat

urda

y or

Sun

day)

26%

27%

29%

22%

42%

35%

27%

33%

31%

44%

– de

sire

d re

duct

ion

in w

orki

ng ti

me

(ave

rage

, in

hour

s)7

3.3

3.4

5.2

0.8

7.1

4.2

5.3

6.8

6.0

8.7

– C

ompa

tibili

ty b

etw

een

wor

king

tim

e an

d fa

mily

life

is a

pro

blem

816

%20

%20

%26

%31

%12

%19

%21

%38

%24

%

With

voc

atio

nal

degr

eeW

ith u

niv.

or

post

-se

cond

ary

tech

-ni

cal c

olle

ge d

egre

e

With

voc

atio

nal

degr

eeW

ith u

nive

rsity

or

[form

er G

DR

] te

ch-

nica

l col

lege

deg

ree

Felix Büchel

546

With

voc

atio

nal

degr

eeW

ith v

ocat

iona

lde

gree

With

uni

vers

ity o

r[fo

rmer

GD

R]

tech

-ni

cal c

olle

ge d

egre

e

Tabl

e 1(

cont

inue

d)Jo

b ch

arac

teris

tics

Wes

t Ger

man

yE

ast G

erm

any

(With

out

(With

out

any

vo-

any

vo-

catio

nal

over

-ad

e-ov

er-

ade-

catio

nal

over

-ad

e-ov

er-

ade-

degr

ee)

qual

ified

quat

equ

alifi

edqu

ate

degr

ee)

qual

ified

quat

equ

alifi

edqu

ate

Pay

men

t of o

vert

ime:

3

– pa

id31

%41

%20

%12

%5%

28%

27%

21%

15%

5%–

time

off i

n lie

u21

%20

%39

%17

%27

%13

%20

%38

%51

%30

%–

part

ly p

aid/

part

ly g

iven

tim

e of

f in

lieu

10%

10%

17%

5%10

%13

%12

%19

%11

%15

%–

no c

ompe

nsat

ion

give

n6%

4%11

%18

%46

%11

%5%

8%18

%40

%–

do n

ot w

ork

over

time

32%

25%

13%

49%

11%

34%

35%

14%

7%10

%S

ecto

rs:

– ag

ricul

ture

/fish

erie

s/m

inin

g/en

ergy

3%4%

4%2%

2%5%

2%7%

5%6%

– ch

emic

al in

dust

ry5%

2%3%

5%4%

2%1%

2%1%

2%–

cons

truc

tion,

qua

rryi

ng, m

iner

als

8%5%

7%2%

3%15

%12

%18

%8%

8%–

com

mer

ce, b

anki

ng, i

nsur

ance

15%

19%

19%

20%

4%13

%15

%14

%23

%6%

– m

etal

wor

king

, ele

ctric

al e

ngin

eerin

g, a

utom

obile

indu

stry

14%

12%

16%

4%14

%5%

8%10

%6%

7%–

publ

ic s

ervi

ce, t

rans

port

, tou

rism

23%

19%

27%

38%

53%

26%

30%

25%

31%

54%

– ot

her

32%

39%

25%

28%

20%

35%

31%

24%

26%

18%

Siz

e of

ent

erpr

ise:

– sm

all e

nter

pris

e (le

ss th

an 2

0 em

ploy

ees)

31%

28%

27%

40%

13%

33%

38%

32%

39%

19%

– m

ediu

m-s

ized

ent

erpr

ise

(bet

wee

n 20

and

200

em

ploy

ees)

28%

28%

25%

34%

25%

43%

37%

33%

22%

37%

– m

ediu

m-s

ized

ent

erpr

ise

(bet

wee

n 20

0 an

d 2

000

empl

oyee

s)22

%28

%22

%10

%22

%12

%15

%18

%24

%21

%–

larg

e en

terp

rise

(2 0

00 e

mpl

oyee

s or

mor

e)18

%16

%27

%16

%41

%12

%10

%17

%15

%23

%Le

ngth

of s

ervi

ce:

– av

erag

e (in

yea

rs)

10.2

8.0

11.8

6.1

11.5

4.2

4.2

7.9

5.7

10.1

Dur

atio

n of

em

ploy

men

t:–

tem

pora

ry e

mpl

oym

ent9

6%7%

3%23

%9%

24%

26%

7%6%

11%

Job

setu

p:–

job-

shar

ing

9%8%

6%2%

1%1%

12%

5%3%

2%Jo

urne

y fr

om h

ome

to w

ork:

– lo

ng jo

urne

y to

wor

k (a

vera

ge, i

n km

)1011

.08.

714

.311

.618

.535

.29.

012

.29.

712

.5–

leng

th o

f jou

rney

(av

erag

e, in

min

utes

)1022

2423

2127

4024

3126

30–

curr

ent j

ob n

ot in

vic

inity

of h

ome

41%

41%

50%

46%

49%

37%

37%

40%

38%

28%

– va

riabl

e pl

ace

of e

mpl

oym

ent

8%12

%7%

12%

7%12

%8%

12%

10%

5%–

jour

ney

to w

ork

is a

gre

at h

ards

hip:

in

finan

cial

term

s12

%5%

10%

3%11

%14

%13

%11

%10

%9%

– jo

urne

y to

wor

k is

a g

reat

har

dshi

p: i

n te

rms

of ti

me

11%

8%10

%11

%15

%9%

8%11

%12

%13

%–

jour

ney

to w

ork

is a

gre

at h

ards

hip:

in

phys

ical

/men

tal t

erm

s8%

2%7%

7%10

%6%

2%6%

8%7%

With

uni

v. o

r po

st-

seco

ndar

y te

ch-

nica

l col

lege

deg

ree


547

With

voc

atio

nal

degr

eeW

ith v

ocat

iona

lde

gree

With

uni

vers

ity o

r[fo

rmer

GD

R]

tech

-ni

cal c

olle

ge d

egre

e

With

uni

v. o

r po

st-

seco

ndar

y te

ch-

nica

l col

lege

deg

ree

Tabl

e 1(

cont

inue

d)Jo

b ch

arac

teris

tics

Wes

t Ger

man

yE

ast G

erm

any

(With

out

(With

out

any

vo-

any

vo-

catio

nal

over

-ad

e-ov

er-

ade-

catio

nal

over

-ad

e-ov

er-

ade-

degr

ee)

qual

ified

quat

equ

alifi

edqu

ate

degr

ee)

qual

ified

quat

equ

alifi

edqu

ate

Wor

k re

quire

men

ts:

(in

eac

h ca

se:

„ful

ly a

pplie

s“)

– va

riety

in th

e jo

b45

%39

%66

%57

%78

%52

%47

%67

%64

%85

%–

heav

y ph

ysic

al w

ork

19%

23%

14%

3%0%

37%

38%

23%

15%

2%–

self-

dete

rmin

ed s

truc

ture

of t

he w

orki

ng d

ay31

%35

%42

%35

%52

%13

%24

%30

%37

%44

%–

wor

king

tim

e de

pend

s on

the

amou

nt o

f wor

k24

%33

%32

%33

%37

%32

%30

%31

%32

%33

%–

stric

t mon

itorin

g of

wor

k pe

rfor

man

ce13

%10

%11

%1%

5%16

%21

%24

%20

%14

%–

rota

ting

wor

k sh

ifts

14%

17%

13%

2%1%

22%

15%

19%

16%

5%–

ofte

n ge

t ang

ry/h

ave

conf

licts

with

hie

rarc

hica

l sup

erio

r1120

%14

%24

%18

%24

%24

%21

%26

%21

%27

%–

good

rel

atio

nshi

p w

ith w

ork

colle

ague

s75

%66

%76

%69

%78

%65

%76

%76

%76

%72

%–

pay

code

term

inat

ion/

prom

otio

n of

the

wor

kfor

ce6%

6%12

%11

%22

%3%

3%6%

8%19

%–

can

alw

ays

unde

rtak

e tr

aini

ng to

adv

ance

occ

upat

iona

l ski

lls19

%13

%39

%38

%54

%35

%20

%42

%40

%56

%–

expo

sed

to n

oxio

us s

ubst

ance

s in

the

envi

ronm

ent

25%

26%

19%

3%3%

35%

39%

26%

15%

6%–

high

men

tal e

ffort

24%

21%

25%

39%

47%

16%

15%

24%

39%

59%

– in

crea

sed

risk

of a

ccid

ents

at w

ork11

39%

43%

38%

23%

10%

63%

59%

49%

33%

27%

Sub

ject

ive

estim

atio

ns:

– sa

tisfa

ctio

n w

ith w

ork12

6.8

6.7

7.1

5.9

7.0

6.6

6.3

6.9

6.6

7.0

– w

illin

gnes

s to

take

on

extr

a, u

npai

d w

ork13

43%

43%

45%

47%

65%

39%

41%

48%

53%

75%

– gr

eat c

once

rn o

ver

safe

ty in

the

wor

kpla

ce9%

12%

9%2%

5%23

%37

%23

%17

%24

%–

easy

to fi

nd a

sim

ilar

job14

26%

24%

28%

45%

21%

26%

8%19

%12

%11

%N

umbe

r of

res

pond

ents

(N

= m

axim

um, u

nwei

ghte

d)1,

080

540

2,00

877

416

6226

590

213

733

0

Tota

l41

1216

99

All

perc

enta

ges:

col

umn

perc

enta

ges

(100

% d

iver

genc

e ro

unde

d up

to

the

next

fig

ure)

. All

resu

ltsw

eigh

ted.

1)P

oten

tial g

ross

mon

thly

sal

ary.

In

the

case

of

part

-tim

e an

d m

inim

um-t

ime

empl

oyee

s, t

he m

onth

lyeq

uiva

lent

, ca

lcul

ated

ove

r an

agr

eed

wor

king

per

iod.

Exc

ludi

ng m

inim

um-t

ime

empl

oyee

s w

orki

ngle

ss t

han

five

hour

s pe

r w

eek

and

the

self-

empl

oyed

.2)

Exc

ludi

ng m

inim

um-t

ime

empl

oyee

s w

orki

ng le

ss t

han

five

hour

s pe

r w

eek

and

the

self-

empl

oyed

.A10

63)

Dep

ende

nt e

mpl

oyee

s =

100

per

cent

.4)

Pay

det

erm

ined

by

leng

th o

f se

rvic

e (a

s w

ell a

s pe

rfor

man

ce)

"on

a re

gula

r ba

sis"

or

"som

etim

es".

5)E

xclu

ding

the

"do

n't

know

"s =

100

per

cent

.6)

Incl

udin

g se

lf-em

ploy

ed f

arm

ers

and

mem

bers

of

thei

r fa

mily

wor

king

on

the

farm

.7)

The

que

stio

n w

as:

"If

you

coul

d ch

oose

the

num

ber

of h

ours

you

wor

ked,

tak

ing

into

acc

ount

tha

t yo

ursa

lary

wou

ld b

e co

mm

ensu

rate

to

the

num

ber

of h

ours

wor

ked,

how

man

y ho

urs

a w

eek

wou

ld y

oupr

efer

to

wor

k?"

(Com

pare

the

ir a

nsw

ers

with

the

ir a

ctua

l wor

king

tim

es.)

8)T

he q

uest

ion

was

: "It

is n

ot a

lway

s so

eas

y fo

r wor

kers

to re

conc

ile w

orki

ng ti

me

with

fam

ily re

spon

sibi

litie

san

d ho

useh

old

task

s. D

o yo

u fin

d th

at t

o be

a p

robl

em?"

. The

res

pond

ents

ans

wer

ed:

"yes

".

9)D

epen

dent

em

ploy

ees

= 1

00 p

erce

nt.

10)

Em

ploy

ees

with

a f

ixed

wor

kpla

ce =

100

per

cent

.11

)T

he r

espo

nden

ts a

nsw

ered

: "f

ully

app

lies"

or

"par

tly a

pplie

s".

12)

On

a sc

ale

from

0 (

"tot

ally

uns

atis

fied”

) to

10

("to

tally

sat

isfie

d”).

13)

The

que

stio

n w

as:

„Wou

ld y

ou p

ut in

ext

ra h

ours

to

finis

h of

f a

task

you

had

alr

eady

beg

un,

even

if it

mea

nt w

orki

ng f

or a

noth

er h

our

and

wou

ld n

ot b

e pa

id a

s ov

ertim

e?“

The

res

pond

ents

ans

wer

ed:

„yes

, of

cou

rse“

.14

)T

he q

uest

ion

was

: „I

f yo

u w

ere

to lo

se y

our

curr

ent

job,

wou

ld it

be

easy

, di

fficu

lt or

pra

ctic

ally

impo

ssib

le f

or y

ou t

o fin

d at

leas

t on

e ot

her

job

that

was

equ

ival

ent?

“ T

he r

espo

nden

ts a

nsw

ered

:„e

asy“

.

Onl

y em

ploy

ees

belo

w t

he a

ge o

f 65

. E

xclu

ding

inte

rnee

s, p

erso

ns r

ecei

ving

tra

inin

g or

fur

ther

tra

inin

g,an

d m

igra

nts

betw

een

Eas

t an

d W

est

Ger

man

y si

nce

1989

. W

est

Ger

man

y:ex

clud

ing

imm

igra

nts

incl

uded

in t

he "

D"

sam

ple

of t

he G

erm

an S

ocio

-Eco

nom

ic P

anel

(G

SO

EP

).

Sou

rce :

ow

n ev

alua

tion

carr

ied

out

by t

he G

erm

an S

ocio

-Eco

nom

ic P

anel

(G

SO

EP

).

Felix Büchel

548

Table 2: Determinants of the probability of possessing subjectively limited problem-solvingcapacity , depending on employment status and other personal characteristics (W est Germany andEast Germany , 1995, probit model)

Covariates West Germany East GermanyCoefficient (Std.dev .) Average Coefficient (Std.dev .) Average

Constant 0.031 (0.330) – 0.830+ (0.458) –Male 0.074 (0.048) 0.60 0.009 (0.059) 0.51Age (in years) –0.011 (0.016) 39.03 –0.035 (0.023) 40.70Age2/100 0.015 (0.020) 16.45 0.054+ (0.028) 17.66Foreigner 0.102 (0.078) 0.15 . ( . ) .Married1 –0.147** (0.052) 0.65 0.043 (0.075) 0.74Restricted on accountof health reasons2 0.518** (0.053) 0.24 0.249** (0.064) 0.33Hauptschulabschluß/8. Klasse 0.121* (0.054) 0.43 0.247** (0.085) 0.25(Realschulabschluß/10. Klasse)3 . ( . ) . . ( . ) .Abitur4 –0.072 (0.123) 0.04 0.207 (0.175) 0.03(Lehre)5 . ( . ) . . ( . ) .Meister (GDR) –0.754* (0.316) 0.01Berufsfachschule 0.027 (0.084) 0.08Schule des Gesundheitswesens 0.070 (0.132) 0.03Fachschule (FRG) 0.133 (0.085) 0.08Beamtenausbildung 0.036 (0.119) 0.04Sonstiger berufsbildenderAbschluß 0.112 (0.106) 0.08Ingenieur-/Fachschule (GDR) –0.290* (0.118) 0.14Fachhochschule –0.244* (0.122) 0.05Hochschule/Universität6 –0.411** (0.105) 0.10 –0.336** (0.123) 0.11Unemployed/in the hiddenlabour reserve7 0.134 (0.100) 0.08 0.018 (0.112) 0.17(Underemployed) . ( . ) . . ( . ) .Adequately employed –0.222** (0.065) 0.61 –0.223* (0.093) 0.45Number of respondents(unweighted): n = 3,285 1,992Log-Likelihood –2149.8 –1265.4Likelihood-Ratio-Statistic 222.1** 118.4**Dependent variables average(weighted) 0.430 0.631

Dependent variable: subjectively limited problem-solving capacity (1 = yes; 0 = no). For the exact research strategies, see text.Significance level: ** = p < 0.01, * = p < 0.05, + = p < 0.10.

1) Including persons living as a married couple.2) “Slightly“/”greatly“ restricted on account of health reasons in carrying out daily tasks.3) Including Fachabitur.4) Only in combination with nonacademic vocational qualification.5) East Germany: Berufsausbildung/Facharbeiterabschluß; including nonacademic vocational qualifications from 1991.6) East Germany: including Fachhochschulabschlüsse from 1991.7) Hidden labour reserve: persons not having registered as unemployed, but wanting to take up a job "as soon as possible".

Legend for German expressions: see table 0.

Covariates in brackets = reference category. Unweighted averages as model documentation.

Only persons with a vocational qualification from Germany below the age of 65. Excluding internees, persons receiving training orfurther training, and migrants between East and West Germany since 1989. West Germany: excluding immigrants included in the "D"sample of the German Socio-Economic Panel (GSOEP).

Source: own evaluation carried out by the German Socio-Economic Panel (GSOEP).


549

Table 3: Determinants of the probability of possessing high morale, depending on employmentstatus and other personal characteristics (W est Germany and East Germany , 1995, probit model)


Constant 1.188** (0.333) – 0.628 (0.470) –Male –0.059 (0.048) 0.60 –0.061 (0.062) 0.51Age (in years) –0.061** (0.017) 39.03 –0.060* (0.023) 40.70Age2/100 0.071** (0.020) 16.45 0.069* (0.028) 17.66Foreigner 0.063 (0.079) 0.15 . ( . ) .Married1 0.303** (0.052) 0.65 0.129+ (0.079) 0.74Restricted on accountof health reasons2 –0.709** (0.055) 0.24 –0.514** (0.070) 0.33Hauptschulabschluß/8. Klasse –0.067 (0.054) 0.43 0.050 (0.087) 0.25(Realschulabschluß/10. Klasse)3 . ( . ) . . ( . ) .Abitur4 0.014 (0.123) 0.04 –0.073 (0.184) 0.03(Lehrabschluß)5 . ( . ) . . ( . ) .Meister (GDR) 0.352 (0.293) 0.01Berufsfachschule –0.026 (0.084) 0.08Schule des Gesundheitswesens –0.161 (0.133) 0.03Fachschule (FRG) –0.054 (0.086) 0.08Beamtenausbildung 0.113 (0.120) 0.04Sonstiger berufsbildenderAbschluß –0.196+ (0.107) 0.08Ingenieur-/Fachschule (GDR) 0.121 (0.124) 0.14Fachhochschule 0.040 (0.121) 0.05Hochschule/Universität6 0.187+ (0.103) 0.10 0.289* (0.128) 0.11Unemployed/in the hiddenlabour reserve7 –0.343** (0.102) 0.08 –0.292* (0.121) 0.17(Underemployed) . ( . ) . . ( . ) .Adequately employed 0.136* (0.066) 0.61 0.246* (0.097) 0.45Number of respondents(unweighted): n = 3,283 1,997Log-Likelihood –2138.5 –1143.4Likelihood-Ratio-Statistic 296.2** 124.2**Dependent variables average(weighted) 0.492 0.282

Dependent variable: morale (1 = high; 0 = [other]). For the exact research strategies, see text.Significance level: ** = p < 0.01, * = p < 0.05, + = p < 0.10.






Felix Büchel

550

Table 4: Determinants of the probability of people’ s concern about their own economic prospects,depending on employment status and other personal characteristics (W est Germany and EastGermany , 1995, probit model)


Constant –0.177 (0.352) – 0.864 (0.580) –Male 0.048 (0.050) 0.60 –0.009 (0.075) 0.51Age (in years) 0.050** (0.017) 39.03 0.026 (0.029) 40.70Age2/100 –0.081** (0.021) 16.45 –0.043 (0.034) 17.66Foreigner 0.251** (0.088) 0.15 . ( . ) .Married1 –0.008 (0.055) 0.65 0.099 (0.096) 0.74Restricted on accountof health reasons2 0.415** (0.060) 0.24 0.449** (0.090) 0.33Hauptschulabschluß/8. Klasse 0.146** (0.057) 0.43 0.122 (0.112) 0.25(Realschulabschluß/10. Klasse)3 . ( . ) . . ( . ) .Abitur4 –0.136 (0.126) 0.04 0.093 (0.227) 0.03(Lehrabschluß)5 . ( . ) . . ( . ) .Meister (GDR) –0.006 (0.358) 0.01Berufsfachschule 0.008 (0.090) 0.08Schule des Gesundheitswesens 0.184 (0.144) 0.03Fachschule (FRG) –0.089 (0.089) 0.08Beamtenausbildung –0.650** (0.119) 0.04Sonstiger berufsbildenderAbschluß 0.043 (0.117) 0.08Ingenieur-/Fachschule (GDR) –0.331* (0.157) 0.14Fachhochschule –0.529** (0.124) 0.05 . ( . ) .Hochschule/Universität6 –0.668** (0.107) 0.10 –0.635** (0.156) 0.11Unemployed/in the hiddenlabour reserve7 0.431** (0.121) 0.08 0.348* (0.179) 0.17(Underemployed) . ( . ) . . ( . ) .Adequately employed –0.145* (0.072) 0.61 –0.404** (0.128) 0.45Number of respondents(unweighted): n = 3,281 1,990Log-Likelihood –1900.9 –727.7Likelihood-Ratio-Statistic 333.2** 104.8**Dependent variables average(weighted) 0.666 0.869

Dependent variable: great concern about their own economic prospects (1 = yes; 0 = no). For the exact research strategies, see text.Significance level: ** = p < 0.01, * = p < 0.05, + = p < 0.10.







551

Table 5: Determinants of the probability of allegiance to a political party , depending on employmentstatus and other personal characteristics (W est Germany and East Germany , 1995, probit model)


Constant –2.235** (0.357) – –0.154 (0.475) –Male 0.154** (0.050) 0.60 0.165** (0.061) 0.51Age (in years) 0.083** (0.018) 39.03 –0.038+ (0.023) 40.70Age2/100 –0.073** (0.021) 16.45 0.063* (0.028) 17.66Foreigner –0.389** (0.083) 0.15 . ( . ) .Married1 0.131* (0.054) 0.65 –0.039 (0.078) 0.74Restricted on accountof health reasons2 0.070 (0.056) 0.24 0.061 (0.065) 0.33Hauptschulabschluß/8. Klasse –0.066 (0.056) 0.43 –0.081 (0.086) 0.25(Realschulabschluß/10. Klasse)3 . ( . ) . . ( . ) .Abitur4 0.218+ (0.127) 0.04 0.600** (0.177) 0.03(Lehrabschluß)5 . ( . ) . . ( . ) .Meister (GDR) –0.433 (0.343) 0.01Berufsfachschule 0.022 (0.086) 0.08Schule des Gesundheitswesens –0.031 (0.136) 0.03Fachschule (FRG) 0.063 (0.089) 0.08Beamtenausbildung 0.187 (0.125) 0.04Sonstiger berufsbildenderAbschluß –0.383** (0.112) 0.08Ingenieur-/Fachschule (GDR) 0.301* (0.123) 0.14Fachhochschule 0.468** (0.127) 0.05Hochschule/Universität6 0.573** (0.109) 0.10 0.602** (0.127) 0.11Unemployed/in the hiddenlabour reserve7 0.115 (0.104) 0.08 0.113 (0.114) 0.17(Underemployed) . ( . ) . . ( . ) .Adequately employed 0.216** (0.069) 0.61 0.238* (0.097) 0.45Number of respondents(unweighted): n = 3,113 1,849Log-Likelihood –1962.9 –1199.1Likelihood-Ratio-Statistic 377.0** 94.8**Dependent variables average(weighted) 0.601 0.399

Dependent variable: allegiance to a political party (1 = yes; 0 = no). For the exact research strategies, see text.Significance level: ** = p < 0.01, * = p < 0.05, + = p < 0.10.






Felix Büchel

552

Table 6: Types of employment status previously held by underemployed workers, broken down byqualification level and sex (W est Germany and East Germany , 1984–1995 and 1991–1995, in %)

West GermanyAccess from: With vocational With university or Total

degree post-secondary technicalcollege degree

Men Women Total Men Women Total Men Women TotalWork commensuratewith qualifications 052 033 041 059 040 049 053 034 042Unemployment/Hidden labour reserve 029 022 025 013 020 016 026 021 023Other economicinactivity 006 037 024 000 015 008 005 033 021(Full-time)training 013 008 010 028 026 027 016 011 013

Total 100 100 100 100 100 100 100 100 100Number ofrespondents 398 524 922 069 071 140 467 595 1,062

East GermanyAccess from: With vocational With university or Total



Total 100 100 100 100 100 100 100 100 100Number ofrespondents 153 177 330 056 094 150 209 271 480


Weighted frequency. Unweighted number of respondents.



553

Table 7: Determinants of the probability of transition to underemployment, depending on the varioustypes of employment status previously held and other characteristics (W est Germany and EastGermany , 1984–1995 and 1991–1995, probit model)


Constant –0.669* (0.278) – 0.303 (0.472) –Male –0.269** (0.048) 0.58 –0.032 (0.066) 0.51Age (in years) 0.006 (0.014) 37.82 0.058* (0.023) 40.43Age2/100 –0.035* (0.017) 15.83 –0.094** (0.028) 17.79Foreigner 0.063 (0.067) 0.15 . ( . ) .Married1 0.082 (0.054) 0.63 0.115 (0.082) 0.74Health restrictions2 –0.036 (0.052) 0.28 0.033 (0.071) 0.32Hauptschulabschluß/8. Klasse 0.390** (0.052) 0.46 0.181* (0.090) 0.26(Realschulabschluß/10. Klasse)3 . ( . ) . . ( . ) .Abitur4 –0.416** (0.135) 0.04 –0.181 (0.200) 0.03(Lehre)5 . ( . ) . . ( . ) .Meister (GDR) –0.126 (0.145) 0.07Berufsfachschule 0.023 (0.075) 0.09Schule des Gesundheitswesens –0.129 (0.139) 0.03Fachschule (FRG) –0.301** (0.102) 0.08Beamtenausbildung –0.562** (0.175) 0.04Sonst. berufsbildender Abschluß 0.327** (0.098) 0.07Ingenieur-/Fachschule (GDR) 0.354** (0.096) 0.15Fachhochschule 0.618** (0.110) 0.04Hochschule/Universität6 0.122 (0.091) 0.09 0.257* (0.113) 0.10Profess. success is not important7 0.112+ (0.063) 0.16 0.008 (0.109) 0.12Regional unemployment figures8 –0.039** (0.008) 8.41 –0.170** (0.013) 15.27Rural area9 0.028 (0.046) 0.56 0.161* (0.074) 0.73(Access from adequate empl.)10 . ( . ) . . ( . ) .Acc. from (other) econ. inactivity 0.497** (0.065) 0.16 –0.129 (0.167) 0.10Access from unemployment/the hidden labour reserve 0.979** (0.067) 0.10 0.783** (0.075) 0.22Access from (full-time) training 0.546** (0.080) 0.08 0.431** (0.121) 0.08No. of respond. (unweighted): n = 4,982 2,615Log-Likelihood –2144.6 –1008.4Likelihood-Ratio-Statistic 1325.0** 546.6**Depend. var. average (weighted) 0.155 0.173

Dependent variable: 1 = transition, from one observation year to the next, from one of the various types of employment status (previous employment status) to underemployment;0 = retention of employment status (transitions between employment status allowed).Significance level: ** = p < 0.01, * = p < 0.05, + = p < 0.10.

1) Including persons living as a married couple.2) ‘Slightly’/‘greatly’ restricted on account of health reasons in carrying out daily tasks.3) Including Fachabitur.4) Only in combination with nonacademic vocational qualification.5) East Germany: Berufsausbildung/Facharbeiterabschluß; including nonacademic vocational qualifications from 1991.6) East Germany: including Fachhochschulabschlüsse from 1991.7) ‘Professional success" is "not so important" or "quite unimportant", according to German nationals (foreigners were not asked).8) West Germany: at regional level (BfLR data - [Bundesforschungsanstalt fuer Landeskunde und Raumordnung - Federal Research Institute for Regional Studies and Town and

Country Planning]); East Germany: at Land level (relating to place of residence).9) Number of inhabitants in place of residence < 50 000 inhabitants.10) Reference date for access status: survey carried out in the previous year to that in which the person's transition to underemployment was first observed.Legend for German

expressions: see table 0.

Covariates in brackets = reference category. Reference date for encoding the covariates: first year. Unweighted averages as model documentation.

Only persons having held one of the types of employment status examined (previous employment status) for two consecutive years (in the second year, underskilled work, inaddition, is allowed). One observation at the most per person (first occurrence of a possible movement). Only persons with a vocational qualification from Germany below the ageof 65. Excluding internees, persons receiving training or further training, and migrants between East and West Germany since 1989. West Germany: excluding immigrants includedin the "D" sample of the German Socio-Economic Panel (GSOEP).


Felix Büchel

554

Table 8: Exit doors from underemployment, broken down by qualification level and sex (W estGermany and East Germany , 1984–1995 and 1991–1995, in %)

West GermanyAccess from: With vocational With university or Total




East GermanyAccess from: With vocational With university or Total




Only persons having been underemployed in the exit year and observed in one of the types of employment status examined in thefollowing year. Only persons with a vocational qualification from Germany below the age of 65. Excluding internees, personsreceiving training or further training, and migrants between East and West Germany since 1989. West Germany: excludingimmigrants included in the "D" sample of the German Socio-Economic Panel (GSOEP).

Weighted frequency. Unweighted number of respondents.



555

Table 9: Determinants of the probability of transition from underemployment or adequateemployment to unemployment/the hidden labour reserve, depending on exit status and othercharacteristics (W est Germany and East Germany , 1984–1995 and 1991–1995, probit model)


Constant 0.166 (0.289) – 1.314** (0.478) –Male –0.010 (0.050) 0.60 –0.269** (0.063) 0.53Age (in years) –0.085** (0.014) 39.52 0.002 (0.024) 40.08Age2/100 0.094** (0.017) 17.08 –0.001 (0.029) 17.16Foreigner 0.100 (0.076) 0.15 . ( . ) .Married1 –0.070 (0.054) 0.67 –0.026 (0.081) 0.77Restricted on account ofhealth reasons2 0.154** (0.052) 0.29 0.021 (0.070) 0.30Hauptschulabschluß/8. Klasse 0.286** (0.057) 0.47 0.182* (0.086) 0.24(Realschulabschluß/10. Klasse)3 . ( . ) . . ( . ) .Abitur4 –0.162 (0.148) 0.04 0.070 (0.203) 0.02(Lehre)5 . ( . ) . . ( . ) .Meister (GDR) –0.197 (0.128) 0.08Berufsfachschule 0.172* (0.080) 0.09Schule des Gesundheitswesens –0.159 (0.159) 0.03Fachschule (FRG) 0.117 (0.091) 0.08Beamtenausbildung –0.332* (0.159) 0.04Sonstiger berufsbildenderAbschluß 0.340** (0.101) 0.09Ingenieur-/Fachschule (GDR) –0.359** (0.098) 0.17Fachhochschule –0.148 (0.149) 0.04Hochschule/Universität6 –0.097 (0.110) 0.09 –0.465** (0.126) 0.11Profess. success is not important7 0.103 (0.066) 0.15 –0.008 (0.106) 0.09Regional unemployment figures8 0.027** (0.008) 8.33 –0.148** (0.012) 14.86Rural area9 –0.015 (0.049) 0.57 0.039 (0.071) 0.72Exit status: underemployment10 0.120* (0.055) 0.24 0.367** (0.069) 0.25Number of respondents(unweighted): n = 4,939 2,404Log-Likelihood –1828.5 1053.1Likelihood-Ratio-Statistic 634.2** 335.2**Dependent variables average(weighted) 0.115 0.212

Dependent variable: 1 = transition, from one observation year to the next, from underemployment or adequate employment to unemployment/the hidden labour reserve; 0 = (other).Significance level: ** = p < 0.01, * = p < 0.05, + = p < 0.10.

1) Including persons living as a married couple.2) "Slightly“/"greatly“ restricted on account of health reasons in carrying out daily tasks.3) Including Fachabitur.4) Only in combination with nonacademic vocational qualification.5) East Germany: Berufsausbildung/Facharbeiterabschluß; including nonacademic vocational qualifications from 1991.6) East Germany: including Fachhochschulabschlüsse from 1991.7) "Professional success" is "not so important" or "quite unimportant", according to German nationals (foreigners were not asked).8) West Germany: at regional level (BfLR data - [Bundesforschungsanstalt fuer Landeskunde und Raumordnung - Federal Research Institute for Regional Studies and Town and

Country Planning]); East Germany: at Land level (relating to place of residence).9) Number of inhabitants in place of residence < 50 000 inhabitants.10) Alternative status: adequate employment.



Only persons having been underemployed or adequately employed in one observation year and interviewed the following year. One observation at the most per person (first occurrenceof a possible movement). Only persons with a vocational qualification from Germany below the age of 65. Excluding internees, pe rsons receiving training or further training, andmigrants between East and West Germany since 1989. West Germany: excluding immigrants included in the "D" sample of the German Socio-Economic Panel (GSOEP).


Felix Büchel

556

Table 10: Determinants of the probability of transition from adequate employment within five years(three years for the former East Germany) to unemployment/the hidden labour reserve, dependingon a possible interim phase of underemployment and other characteristics (W est Germany and EastGermany , 1984–1995 and 1991–1995, probit model)


Constant 0.508 (0.393) – –0.237 (0.588) –Male –0.284** (0.071) 0.64 –0.279** (0.078) 0.56Age (in years) –0.096** (0.022) 35.65 –0.065* (0.028) 39.29Age2/100 0.111** (0.028) 14.00 0.097** (0.035) 16.49Foreigner –0.029 (0.138) 0.09 . ( . ) .Married1 0.031 (0.080) 0.63 0.068 (0.106) 0.80Restricted on account ofhealth reasons2 0.143+ (0.080) 0.23 0.122 (0.088) 0.25Hauptschulabschluß/8. Klasse 0.189* (0.077) 0.47 0.106 (0.109) 0.23(Realschulabschluß/10. Klasse)3 . ( . ) . . ( . ) .Abitur4 –0.087 (0.182) 0.04 –0.113 (0.246) 0.03(Lehre)5 . ( . ) . . ( . ) .Meister (GDR) –0.113 (0.138) 0.09Berufsfachschule 0.085 (0.112) 0.09Schule des Gesundheitswesens –0.244 (0.211) 0.03Fachschule (FRG) 0.030 (0.125) 0.08Beamtenausbildung –0.548** (0.208) 0.05Sonst. berufsbildender Abschluß 0.562** (0.170) 0.05Ingenieur-/Fachschule (GDR) –0.320** (0.121) 0.17Fachhochschule –0.003 (0.211) 0.04Hochschule/Universität6 –0.008 (0.135) 0.10 –0.359* (0.144) 0.12Profess. success is not important7 0.292** (0.088) 0.14 0.449+ (0.140) 0.06Regional unemployment figures8 0.012 (0.010) 8.96 0.032 (0.019) 11.00Rural area9 0.018 (0.068) 0.55 0.112 (0.087) 0.71Interim phase: underemploymt.10 –0.856** (0.134) 0.15 –0.307* (0.123) 0.13No. of respond. (unweighted): n = 2,911 1,536Log-Likelihood –894.0 –702.5Likelihood-Ratio-Statistic 223.0** 57.2**Dependent variables average(weighted) 0.106 0.215

Dependent variable: 1 = transition within five years (three years for the former East Germany) from adequate employment to unemployment/the hidden labour reserve; 0 = (other).Significance level: ** = p < 0.01, * = p < 0.05, + = p < 0.10.


Country Planning]); East Germany: at Land level (relating to place of residence).9) Number of inhabitants in place of residence < 50 000 inhabitants.10) Transition to unemployment/the hidden labour reserve following a phase of underemployment.



Only persons having been in adequate employment in one of the observation years (1984–1995) (East Germany: 1991–1995) and interviewed within the following five years (threeyears for the former East Germany). One observation at the most per person (first occurrence of a possible movement). Only persons with a vocational qualification from Germanybelow the age of 65. Excluding internees, persons receiving training or further training, and migrants between East and West Germany since 1989. West Germany: excludingimmigrants included in the "D" sample of the German Socio-Economic Panel (GSOEP).



557

Table 11: Determinants of the probability of transition from underemployment or unemployment/thehidden labour reserve to adequate employment, depending on exit status and other characteristics(West Germany and East Germany , 1984–1995 and 1991–1995, probit model)


Constant –0.409 (0.357) – –0.023 (0.611) –Male 0.239** (0.059) 0.52 0.292** (0.086) 0.44Age (in years) 0.057** (0.018) 38.96 0.104** (0.030) 40.91Age2/100 –0.092** (0.023) 16.73 –0.152** (0.037) 18.15Foreigner –0.444** (0.088) 0.24 . ( . ) .Married1 –0.085 (0.067) 0.67 0.304** (0.109) 0.75Restricted on account ofhealth reasons2 –0.142* (0.065) 0.33 –0.263** (0.094) 0.35Hauptschulabschluß/8. Klasse –0.169* (0.072) 0.54 –0.212+ (0.115) 0.33(Realschulabschluß/10. Klasse)3 . ( . ) . . ( . ) .Abitur4 0.156 (0.224) 0.02 –0.033 (0.273) 0.02(Lehre)5 . ( . ) . . ( . ) .Meister (GDR) 0.414* (0.198) 0.04Berufsfachschule 0.038 (0.097) 0.09Schule des Gesundheitswesens 0.141 (0.206) 0.02Fachschule (FRG) 0.157 (0.141) 0.04Beamtenausbildung 0.397 (0.246) 0.01Sonst. berufsbildender Abschluß –0.469** (0.122) 0.16Ingenieur-/Fachschule (GDR) –0.122 (0.128) 0.15Fachhochschule 0.017 (0.149) 0.05Hochschule/Universität6 0.118 (0.135) 0.05 –0.116 (0.158) 0.08Professional successis not important7 –0.345** (0.082) 0.17 –0.371** (0.137) 0.15Regional unemployment figures8 –0.037** (0.010) 8.28 –0.142** (0.016) 14.96Rural area9 –0.103+ (0.060) 0.56 –0.346** (0.095) 0.76Exit status: underemployment –0.358** (0.059) 0.61 –0.261** (0.086) 0.43Number of respondents(unweighted): n = 2,375 1,320Log-Likelihood –1305.3 –608.0Likelihood-Ratio-Statistic 584.6** 263.4**Dependent variables average(weighted) 0.301 0.226

Dependent variable: 1 = transition, from one observation year to the next, from underemployment or unemployment/the hidden labour reserve to adequate employment; 0 = (other).

Significance level: ** = p < 0.01, * = p < 0.05, + = p < 0.10.

1) Including persons living as a married couple.2) ‘Slightly’/‘greatly’ restricted on account of health reasons in carrying out daily tasks.3) Including Fachabitur.4) Only in combination with nonacademic vocational qualification.5) East Germany: Berufsausbildung/Facharbeiterabschluß; including nonacademic vocational qualifications from 1991.6) East Germany: including Fachhochschulabschlüsse from 1991.7) ‘Professional success’ is ‘not so important’ or ‘quite unimportant’, according to German nationals (foreigners were not asked).8) West Germany: at regional level (BfLR data - [Bundesforschungsanstalt fuer Landeskunde und Raumordnung - Federal Research Institute for Regional Studies and Town and

Country Planning]); East Germany: at Land level (relating to place of residence).9) Number of inhabitants in place of residence < 50 000 inhabitants.10) Alternative status: unemployment/the hidden labour reserve.



Only persons having been underemployed or unemployed/in the hidden labour reserve in one observation year and interviewed the following year. One observation at the most perperson (first occurrence of a possible movement). Only persons with a vocational qualification from Germany below the age of 65. Excluding internees, persons receiving trainingor further training, and migrants between East and West Germany since 1989. West Germany: excluding immigrants included in the "D" sample of the German Socio-EconomicPanel (GSOEP).


Felix Büchel

558

Table 12: Determinants of the probability of transition from unemployment/the hidden labourreserve within five years (three years for the former East Germany) to adequate employment,depending on a possible interim phase of underemployment and other characteristics (W estGermany and East Germany , 1984–1995 and 1991–1995, probit model)


Constant –0.948 (0.674) – –3.253** (1.002) –Male 0.640** (0.122) 0.46 0.401** (0.152) 0.41Age (in years) 0.142** (0.037) 35.91 0.258** (0.052) 41.85Age2/100 –0.222** (0.047) 14.44 –0.350** (0.064) 19.03Foreigner –0.676** (0.196) 0.20 . ( . ) .Married1 –0.247+ (0.132) 0.61 –0.038 (0.185) 0.78Restricted on account ofhealth reasons2 –0.265* (0.133) 0.33 –0.465** (0.179) 0.30Hauptschulabschluß/8. Klasse –0.232+ (0.143) 0.53 –0.380* (0.197) 0.38(Realschulabschluß/10. Klasse)3 . ( . ) . . ( . ) .Abitur4 –0.197 (0.346) 0.03 0.212 (0.422) 0.02(Lehre)5 . ( . ) . . ( . ) .Meister (GDR) 0.291 (0.338) 0.04Berufsfachschule 0.093 (0.193) 0.10Schule des Gesundheitswesens 0.372 (0.377) 0.02Fachschule (FRG) 0.471+ (0.256) 0.06Beamtenausbildung 0.445 (0.465) 0.02Sonst. berufsbildender Abschluß –0.473+ (0.274) 0.11Ingenieur-/Fachschule (GDR) –0.057 (0.241) 0.13Fachhochschule –0.445 (0.405) 0.03Hochschule/Universität6 –0.385 (0.253) 0.06 –0.017 (0.303) 0.07Profess. success is not important7 –0.541** (0.146) 0.21 –0.093 (0.217) 0.17Regional unemployment figures8 –0.036* (0.018) 9.51 –0.063* (0.029) 12.54Rural area9 –0.055 (0.120) 0.53 –0.371* (0.162) 0.74Interim phase:underemployment10 –1.582** (0.133) 0.35 –1.409** (0.182) 0.27Number of respondents(unweighted): n = 707 463Log-Likelihood –317.7 –212.3Likelihood-Ratio-Statistic 373.0** 163.2**Dependent variables average(weighted) 0.456 0.307

Dependent variable: 1 = transition within five years (three years for the former East Germany) from unemployment/the hidden labour reserve to adequate employment; 0 = (other).Significance level: ** = p < 0.01, * = p < 0.05, + = p < 0.10.


Country Planning]); East Germany: at Land level (relating to place of residence).9) Number of inhabitants in place of residence < 50 000 inhabitants.10) Transition from unemployment/the hidden labour reserve following a phase of underemployment.

Legend for German expressions: see table 0.Covariates in brackets = reference category. Reference date for encoding the covariates: first year. Unweighted averages as model documentation.

Only persons having been unemployed/in the hidden labour reserve in one of the observation years (1984–1995) (East Germany: 1991–1995) and interviewed within the followingfive years (three years for the former East Germany). One observation at the most per person (first occurrence of a possible transition). Only persons with a vocational qualificationfrom Germany below the age of 65. Excluding internees, persons receiving training or further training, and migrants between East and West Germany since 1989. West Germany:excluding immigrants included in the "D" sample of the German Socio-Economic Panel (GSOEP).



559

Table 13: Determinants of income growth, depending on adequate employment at the beginning ofthe observation period and other influential factors (W est Germany and East Germany , 1984–1995and 1991–1995, OLS)


Constant 0.449** (0.064) – 0.418** (0.134) –Male 0.070** (0.009) 0.61 –0.015 (0.013) 0.54Age (in years) 0.005 (0.003) 40.04 0.001 (0.005) 40.64Age2/100 –0.008* (0.003) 17.42 –0.0002 (0.006) 17.50Foreigner –0.044** (0.014) 0.16 . ( . ) .Married1 0.013 (0.009) 0.67 0.009 (0.016) 0.77Restr. on acc. of health reasons2 –0.024 (0.009) 0.28 –0.027+ (0.014) 0.28Hauptschulabschluß/8. Klasse –0.073* (0.010) 0.46 –0.064** (0.018) 0.20(Realschulabschluß/10. Klasse)3 . ( . ) . . ( . ) .Abitur4 0.081** (0.021) 0.04 0.059 (0.037) 0.03(Lehre)5 . ( . ) . . ( . ) .Meister (GDR) 0.009 (0.026) 0.07Berufsfachschule 0.031* (0.015) 0.08Schule des Gesundheitswesens –0.014 (0.025) 0.03Fachschule (FRG) 0.074** (0.017) 0.08Beamtenausbildung –0.023 (0.020) 0.05Sonst. berufsbildender Abschluß –0.067** (0.019) 0.09Ingenieur-/Fachschule (GDR) 0.127** (0.018) 0.19Fachhochschule 0.193** (0.023) 0.04Hochschule/Universität6 0.271** (0.018) 0.08 0.266** (0.022) 0.12Profess. success is not important7 –0.021+ (0.012) 0.13 0.010 (0.025) 0.06Regional unemployment figures8 –0.003* (0.001) 8.35 0.001 (0.005) 15.12Rural area9 0.001 (0.008) 0.57 –0.056** (0.014) 0.70Initial income/10010 –0.014** (0.0005) 29.21 –0.022** (0.001) 18.24Observation period (in years) 0.041** (0.001) 6.27 0.140** (0.006) 3.06Number of times the personhas switched companies11 0.014** (0.003) 0.92 –0.036** (0.007) 0.68No. of years of econ. inactivity12 –0.037** (0.007) 0.10 –0.045 (0.050) 0.01Number of years the personcould not be observed13 0.016 (0.011) 0.09 0.009 (0.032) 0.04Number of years unemployed/in the hidden labour reserve –0.066** (0.009) 0.11 –0.054** (0.015) 0.12At the beginning of the incomeobservation period:in an underskilled job14 –0.049** (0.011) 0.21 –0.056** (0.016) 0.18No. of respond. (unweighted): n = 3,396 1,625R2adj. .45 .47F value 113.8** 73.6**Depend. var. average (weighted) 0.3556 0.4895

Dependent variable: plotted on a logarithmic scale (final income/initial income) covering the longest possible observation period; “Income”: potential gross monthly salary, basedon information relating to working time (full-time equivalent). Significance level: ** = p < 0.01, * = p < 0.05, + = p < 0.10.Footnotes 1 to 9: see Table 1210) Gross monthly salary in DM (full-time equivalent). 11) During the observation period. 12) "Number of years" below: number of years the person held the correspondingemployment status during the observation period. 13) Years missing from the panel records (non-response unit). 14) Research strategies: see text. Reference category: adequateemployment.Legend for German expressions: see table 0. Covariates in brackets = reference category. Unweighted averages as model documentation. Reference date for encoding thecovariates: final year. Full-time equivalents on the basis of a 40-hour working week. Only persons with a vocational qualification from Germany below the age of 65. Excludinginternees, persons receiving training or further training, and migrants between East and West Germany since 1989. West Germany: excluding immigrants included in the "D"sample of the German Socio-Economic Panel (GSOEP).Source: own evaluation carried out by the German Socio-Economic Panel (GSOEP).

Felix Büchel

560

Table 14: Determinants of income growth among adequately employed people, depending onpossible interim phases of underemployment and other influential factors (W est Germany and EastGermany , 1984–1995 and 1991–1995, OLS)


Constant 0.496** (0.065) – 0.440** (0.144) –Male 0.056** (0.010) 0.63 –0.026+ (0.014) 0.55Age (in years) –0.000 (0.003) 39.56 –0.000 (0.006) 40.58Age2/100 –0.002 (0.003) 17.03 0.003 (0.007) 17.44Foreigner –0.047** (0.015) 0.11 . ( . ) .Married1 0.014 (0.010) 0.66 0.004 (0.018) 0.77Restricted on account ofhealth reasons2 –0.040** (0.010) 0.27 –0.021 (0.015) 0.27Hauptschulabschluß/8. Klasse –0.070** (0.010) 0.45 –0.062** (0.020) 0.19(Realschulabschluß/10. Klasse)3 . ( . ) . . ( . ) .Abitur4 0.074** (0.020) 0.05 0.047 (0.038) 0.03(Lehre)5 . ( . ) . . ( . ) .Meister (GDR) –0.024 (0.026) 0.08Berufsfachschule 0.023 (0.016) 0.09Schule des Gesundheitswesens –0.020 (0.024) 0.03Fachschule (FRG) 0.045** (0.016) 0.09Beamtenausbildung –0.034+ (0.019) 0.06Sonst. berufsbildender Abschluß –0.042+ (0.023) 0.05Ingenieur-/Fachschule (GDR) 0.167** (0.020) 0.17Fachhochschule 0.177** (0.024) 0.04Hochschule/Universität6 0.225** (0.018) 0.09 0.270** (0.023) 0.12Profess. success is not important7 –0.005 (0.013) 0.12 –0.026 (0.030) 0.05Regional unemployment figures8 –0.004** (0.001) 8.36 –0.000 (0.005) 15.11Rural area9 0.008 (0.009) 0.57 –0.041** (0.015) 0.69Initial income/10010 –0.012** (0.0005) 30.61 –0.020** (0.001) 18.85Observation period (in years) 0.045** (0.001) 5.96 0.145** (0.006) 3.01Number of times the personhas switched companies11 0.020** (0.003) 0.86 –0.030** (0.008) 0.59No. of years of econ. inactivity12 –0.041** (0.011) 0.07 –0.070 (0.065) 0.01Number of years the personcould not be observed13 0.016 (0.012) 0.08 0.033 (0.038) 0.03Number of years unemployed/in the hidden labour reserve –0.071** (0.013) 0.08 –0.051* (0.021) 0.08No. of years underemployed14 –0.018+ (0.011) 0.09 –0.055+ (0.033) 0.03No. of respond. (unweighted): n = 2,835 1,295R2adj. .48 .50F value 107.5** 65.5**Depend. var. average (weighted) 0.3505 0.4917

Dependent variable: plotted on a logarithmic scale (final income/initial income) covering the longest possible observation period; “Income”: potential gross monthly salary,based on information relating to working time (full-time equivalent).

Significance level: ** = p < 0.01, * = p < 0.05, + = p < 0.10.

Footnotes 1 to 9: see Table 1210) Gross monthly salary in DM (full-time equivalent).11) During the observation period.12) "Number of years" below: number of years the person held the corresponding employment status during the observation period.13) Years missing from the panel records (non-response unit).14) Research strategies: see text.


Covariates in brackets = reference category. Unweighted averages as model documentation. Reference date for encoding the covariates: final year. Full-time equivalents on thebasis of a 40-hour working week. Only persons adequately employed from the beginning to the end of the examination period.

Only persons with a vocational qualification from Germany below the age of 65. Excluding internees, persons receiving training or further training, and migrants between Eastand West Germany since 1989. West Germany: excluding immigrants included in the "D" sample of the German Socio-Economic Panel (GSOEP).
