Modelling reallocation processes in long-term labour market projections

Tobias Maier; Caroline Neuber-Pohl; Anke Mönnig; Gerd Zika; Michael Kalinowski

doi:10.1007/s12651-017-0220-x

Modelling reallocation processes in long-term labour market projections

Maier, Tobias; Neuber-Pohl, Caroline; Mönnig, Anke; Zika, Gerd; Kalinowski, Michael 2017-04-05 00:00:00 J Labour Market Res (2017) 50:67–90 DOI 10.1007/s12651-017-0220-x ARTICLE Modelling reallocation processes in long-term labour market projections 1 1 2 3 1 Tobias Maier · Caroline Neuber-Pohl · Anke Mönnig · Gerd Zika · Michael Kalinowski Accepted: 19 January 2017 / Published online: 5 April 2017 © The Author(s) 2017. This article is available at SpringerLink with Open Access. Abstract Long-term labour market projections are a popu- scenario comparisons. Our results suggest that considering lar tool for assessing future skill needs and the possibility of reallocations but also additionally their dynamics has sub- skill shortages. It is often noted that reallocation processes stantial effects on the projection outcomes. They help draw in the German labour market are hindered due to its strong an insightful picture of the future labour market and prevent standardization and occupational segmentation. However, it over- or understating the potential for labour shortages in is possible that persons leave the occupation for which they several occupations. have been trained for. Disregarding such reallocations and We conclude that the assumptions about how realloca- their dynamics in the projection model is likely to distort tions differ by occupation and to what extent they can be the results and lead to inaccurate practical advice. realized by wage impulses is essential for projection results In this article, we describe for the ﬁrst time, how reallo- and their interpretation. Furthermore, we ﬁnd that in the cations in the labour market can be modelled using occu- German labour market, wage adjustments cannot balance pational ﬂexibility matrices and wage dynamics. Here, it is the labour demand and supply for occupations completely. shown that employers react to labour scarcity by increas- ing wages to attract workers who to some extent can adjust Keywords Labour market · Projections · Germany · their mobility behaviour accordingly. We analyse the aggre- Occupational mobility · Education · Wage development gate impact of this implementation of a reallocation process of labour supply on the projection results by the means of JEL I25 · J20· J21·J22·J23· J24·O11· O15 Modellierung von Anpassungsprozessen in Tobias Maier tobias.maier@bibb.de langfristigen Arbeitsmarktprojektionen Caroline Neuber-Pohl neuber-pohl@bibb.de Zusammenfassung Langfristige Arbeitsmarktprojektio- nen stellen ein beliebtes Analyseinstrument dar, um zu- Anke Mönnig künftige Fachkräftebedarfe und -engpässe aufzuzeigen. moennig@gws-os.com Es wird oft angemerkt, dass gerade der stark standardi- Gerd Zika sierte und beruﬂich segmentierte deutsche Arbeitsmarkt gerd.zika@iab.de Reallokationsprozesse von Arbeitsangebot und -bedarf Michael Kalinowski nach Berufen erschwert. Nichtsdestotrotz sind Wechsel kalinowski@bibb.de aus dem erlernten Beruf keine Seltenheit und müssen bei Federal Institute for Vocational Education and Training, einer langfristigen Projektion nach Berufen berücksichtigt Bonn, Germany werden, sofern keine inadäquaten Handlungsempfehlungen Institute of Economic Structures Research, Osnabrueck, aus vermeintlichen Fachkräfteengpässen und -überschüssen Germany abgeleitet werden sollen. Institute for Employment Research, Nuremberg, Germany K 68 T. Maier et al. In diesem Artikel beschreiben wir erstmals, wie die Im- supply hinges on today’s education attainment. Here, the plementierung eines Reallokationsprozesses durch beruﬂi- occupation represents an institutional link between educa- che Flexibilitätsmatrizen und berufsfeldspeziﬁscher Löh- tion and employment (c.f. Weber 1972; Mayer and Carroll ne stattﬁnden kann. So zeigen wir, dass Arbeitgeber auf 1987; Abraham et al. 2011). In such a market, workers Engpässe durch Lohnerhöhungen reagieren, woraufhin Ar- cannot be regarded as homogeneous and perfectly substi- beitnehmer ihr Mobilitätsverhalten anpassen. Anhand von tutable. The production of different goods or services call Szenarien analysieren wir die Auswirkungen unterschiedli- for different specialized skills and, therefore, not every em- cher Annahmen zur Lohnentwicklung in den Berufen und ployee is suited for every job. This is why, for Germany it is deren Effekte auf das Anpassungsverhalten des Arbeitsan- essential to project occupation-speciﬁc labour demand and gebots. Unsere Ergebnisse zeigen, dass sich die Berück- supply in order to yield insightful results (Lapointe et al. sichtigung beruﬂichen Mobilitätsverhaltens sowie eine dy- 2008; CEDEFOP 2012; Helmrich and Zika 2010). namische Entwicklung desselben substanziell in den lang- However, although these submarkets are linked to a spe- fristigen Projektionsergebnissen niederschlagen. Hierdurch ciﬁc occupation, they are not totally restrictive. The trans- ergibt sich ein differenzierteres Bild über mögliche Fach- ferability of task-based human capital enables occupational kräfteengpässe und -überhänge sowie mögliche Handlungs- mobility to related ﬁelds (Gathmann and Schönberg 2010). empfehlungen. In fact, Nisic and Trübswetter (2012) calculate that ev- Als Fazit lässt sich festhalten, dass mögliche Lohnanpas- ery year about 3.4% of Germany’s employed population sungen und damit verbundene Berufswechsel zu einem bes- change their occupation. To put this into perspective, Nisic seren Ausgleich von Arbeitsangebot und -nachfrage nach and Trübswetter (2012) calculate a yearly rate of 10.8% Berufen führen können und dass Annahmen über den Ab- in Great Britain. For Denmark, Groes et al. (2015)talk of lauf dieser Prozesse das Ergebnis stark beeinﬂussen. Zudem a yearly occupational mobility rate of 20% and Moscarini können wir für den deutschen Arbeitsmarkt konkludieren, andThomsson(2007) estimate a monthly rate of 3.5% dass nicht alle theoretischen Engpässe lediglich über Lohn- among male workers in the US. Thus, in the international erhöhungen lösbar sind. comparison, a yearly rate of 3.4% may actually be a rel- atively small number. Nevertheless, this level of mobility can to a certain extent be thought to resolve misallocations 1 Introduction of the working population. Furthermore, disregarding the opportunities and limitations of occupational ﬂexibility and The German economy and labour market are subject to its dynamics in projection models is likely to distort the structural change over time. Demographic change, techno- results (cf. Brücker et al. 2013; Brunow and Garloff 2011). logical progress, and globalisation will frame the behaviour Notwithstanding, projection models have to trade-off of market participants. Political planners have a special in- transparency of results and accuracy to some extent; ac- terest in having some knowledge about the future – be it curately reﬂecting all underlying mechanisms may cause for budgetary planning or preliminary policy assessments. separate effects not to be identiﬁable and results not inter- In addition, regarding future developments of the labour pretable (Wilson 2001). Therefore, the decision of whether market, a concern is whether the supply of skills will suf- or not and how to implement reallocation dynamics in ﬁce the demand of the economy, such that growth can a projection model of the German labour market is not spur, or whether there is a possibility of labour shortages. trivial. Here, long-term labour market projections are a more and Helmrich and Zika (2010) for the ﬁrst time model oc- more popular tool for policy consulting (Wilson 2001). To- cupational ﬂexibilities into a long-term projection of the day many countries have such projections (cf. for example German labour market, the BIBB-IAB qualiﬁcation and CEDEFOP 2009 and 2012 for Europe; Dupuy 2012 for occupational ﬁeld projections (QuBe, henceforth). Based the Netherlands; Gajdos and Zmurkow-Poteralska 2014 for on this, Maier et al. (2014) propose a dynamic realloca- Poland; Bonin et al. 2007; Maier et al. 2014; and Vogler- tion mechanism for the qube model, which redistributes Ludwig and Düll 2013 for Germany; Lapointe et al. 2008 labour supply to labour demand via occupational mobil- for Canada; Lepic and Koucky 2012 for the Czech Repub- ity given wages. This is a novel approach to model long- lic; Lockard and Wolf 2012 for the US; Tiainen 2012 for term projections and to our knowledge has not been done Finland; Papps 2001for New Zealand; UK Commission for in any other labour market projection so far. In the model Employment and Skills 2011 for the UK). of Maier et al. (2014), employers respond to occupation- Especially in Germany, where the labour market is highly speciﬁc labour scarcity by raising wages, which in turn segmented into occupation-speciﬁc submarkets (cf. Mayer causes trained workers and workers from related disciplines and Carroll 1987; Allmendinger 1989; Shavit and Müller to more often offer their work in this occupational submar- 2000;OECD 2003), the balance of the labour demand and ket. In this paper, we wish to highlight the impact of this K Modelling reallocation processes in long-term labour market projections 69 modelling approach on the QuBe projection results and the effect on the projection results. In this section, to start with, overall importance of considering reallocation mechanism we will brieﬂy discuss the choice of a purely wage driven in labour market projections in the context of the evaluation mechanism reﬂecting on related literature on the topic of of possible hazards of labour supply shortages in the future. turnover, employer recruitment strategy, and the drivers of Our analysis will show in which occupations, we can rely occupational mobility in general. on market mechanisms to solve possible labour shortages via wage dynamics and in which occupations, enterprises 2.1 The employer’s adjustment mechanism and policy makers have to intervene by for example im- proving working conditions in general or providing further Projection results are often said to exaggerate the extent educational training. of possible labour shortages in the future. This critique of- In the following, we ﬁrst discuss whether wage-based ten addresses that adjustment mechanisms of employers are dynamics of the reallocation process are adequate by re- neglected in the analysis (cf. for example Brücker et al. viewing recent literature on this topic (Sect. 2). In the 2013). Brunow and Garloff (2011) even reject the idea of third section, we brieﬂy give an intuitive introduction to future labour market shortages in total. They argue that in the QuBe model and describe its reallocation mechanism in the event of a tightening labour market, employers have more detail. In the fourth section, we outline the different plenty of ways to adjust adequately and prevent a shortage data sources used for the QuBe model and how the real- situation. They suggest that ﬁrms will react to the antici- location dynamics where operationalized. Sect. 5 presents pation of a shortage by substituting their labour demands results from scenario comparisons, which illustrate the ef- by automating processes or hiring workers from abroad. fect of this modelling on the projection results. Here, we Also ﬁrms could alter their stock of capital and produce ﬁrst assess the overall impact of implementing the reallo- less, thereby demanding less labour. Brunow and Garloff cation mechanism in QuBe (Sect. 5.1). Then we show how (2011) also highlight the importance of wages, which they the dynamic adjustment of employers and workers to each consider ’upward ﬂexible’ enough to attract the necessary other take a great part in the overall effect (Sect. 5.2). After labour supply. this, we discuss how the interpretation of the results are Economic theory, likewise, predicts a relationship be- strongly inﬂuenced also by the implicitly modelled limita- tween wages and relative labour supply. Especially in the tions of wage dynamics in balancing the labour market by search and matching literature labour market tightness ex- presenting results from wage policy scenarios (Sect. 5.3) plicitly enters the wage equation such that a shortage of and discussing to what extent the calculated optimal ﬂexi- applicants always corresponds to higher wages (cf. for ex- bility of the workforce is achievable via the wage mecha- ample Pissarides 2000). Montgomery (1991), for example, nism (Sect. 5.4). In Sect. 6, we conclude and give an outlook uses a related model set-up to explain wage differences on future research. across industries. Here, ﬁrms who value ﬁlling their va- cancy most, pay the highest wage in order to overcome coordination problems and attract the most applicants to 2 Theoretical assumptions and related empirical their opening. ﬁndings However, Bechmann et al. (2012) show that wage policy may be less important to German recruiters. They analyse In order to account for reallocation dynamics in their pro- data of the IAB Establishment panel , where ﬁrms were jection model Maier et al. (2014) let employer-set wages asked which strategies they used or would use to allevi- partially depend on labour supply scarcity. Labour supply, ate labour shortages. The most important strategy, in fact, in turn, responds to differences in relative wages of occu- seems to be further training of the current workforce, which pations by changing their occupational mobility behaviour was chosen as very important by 42% of the surveyed ﬁrms. in that the workers propensity to stay in their training oc- Next to other means of recruiting from within the company, cupation correlates positively with a lower outside option. as for example later retirements or apprenticeship programs, In this model set-up, wage is the only explicit adjustment also the attractiveness of the job offer was stated to be tar- channel of employers and worker behaviour in response to geted. With 34% of the establishments highlighting its im- misallocations of labour. All other factors, which inﬂuence portance making the offer desirable seems to be the second mobility decisions of workers, are assumed to follow a con- most important strategy of ﬁrms. In contrast, wages seem stant time trend. Other factors, which drive wage setting of to be less important. Only 11% of the ﬁrms consider pay- the employer, are assumed to relate to the production pro- cess and outside wage pressures. The Establishment Panel of the Institute for Employment Research In the following sections, we will describe this mecha- (IAB) representatively surveys about 16,000 German establishments nism in more detail and outline its empirical foundation and on their employment policies and related topics since 1993. K 70 T. Maier et al. ing higher wages as an important strategy. It is, however, voluntary lay-offs are associated with a switch to lower still a strategy for 47% of the surveyed ﬁrms, even though wages (McLaughlin 1991), which Gibbons and Katz (1991) 36% indicate that a main problem concerning recruiting is, explain with the ‘lemon effect’ causing laid-off workers in fact, too high wage demands of applicants (Bechmann having troubles with ﬁnding a new job. The importance et al. 2012). of the nature of the switch is also highlighted by recent Eventually, Dustman and Glitz (2015)and Dustman et al. results of Fitzenberger et al. (2015). Providing evidence (2009) ﬁnd empirical evidence for the impact of the struc- concerning the occupational mobility of recent apprentice- ture of skill supply on wages. Using IAB Establishment ship completers in the German labour market, they ﬁnd that Panel data from 1985 and 1995, Dustman and Glitz (2015) mere job switches inside the occupation but between ﬁrms investigate whether employers in West Germany react to most often lead to a wage loss, while occupational mobil- a change in the skill mix of the workforce by adjusting ity is associated with a wage gain in most cases. However, wages or the production intensity, where they distinguish they point out that occupation-and-ﬁrm switches only result between switching to production of goods, which can be in a gain if this switch reﬂects an occupational upgrading, produced by the skills available, or producing the same while occupation switches within the ﬁrm, which reﬂect goods but adjusting the skill application. They conclude a switch to a better ﬁtting position, are usually associated that ﬁrms adjust mainly by the latter. Concerning wage ad- with awagegain. justments, they ﬁnd that wages are only signiﬁcantly elastic Other research points toward the increasing wage in- with respect to skill supply in the nontradable and manu- equality. Groes et al. (2015) point out that mainly low and facturing sector, where a 1% increase of skill supply corre- high income earners switch occupations and that downward sponds to a 0.4% and 0.1% decrease in wages, respectively. mobility seems to be a phenomenon of low income earners. Dustman et al. (2009) come to a similar conclusion. Taking An explanation for this, according to Groes et al. (2015), is advantage of the change in skill structure of the German that occupations with rising productivities layoff their low labour market induced by the reuniﬁcation, they show that skilled workers (and typically low wage earners), leaving the relative abundance of lower skilled workers after the in- them to seek work in other occupations, while high skilled tegration of the East German Länder increased skill returns. workers move out of the declining productivity occupations To sum up, there is evidence of ﬁrms reacting to labour in order to obtain higher wages. As a result, again only the market tightness by raising wages in order to attract suf- high skilled workers are hypothesized to experience wage ﬁcient applicants to their vacancies. However, the extent increases when switching their occupation. of the wage mechanism may be relatively small as ﬁrms The literature on task biased technological change ex- also use other strategies to overcome recruitment problems. plains the observed trends in wage inequality by job polar- These include training and solutions for better working con- ization. Emerging new technologies, which automate many ditions (cf. Bechmann et al. 2012). routine tasks, and globalisation, which poses new opportu- nities for offshoring (see also Grossman and Rossi-Hans- 2.2 The worker’s adjustment mechanism berg 2008), cause redundancy of domestic labour in some occupations (see for a summary Acemoglu and Autor 2011; In labour economics, there has been a long debate about Goos et al. 2009). Such a trend can also be found for Ger- whether job or occupational mobility is associated with many (cf. Spitz-Oener 2006). Cortes (2016) explains this a wage gain or a penalty. The classic island model by Lu- polarization effect further by the induced sorting on ability cas and Prescott (1974) would predict that negative demand among the workforce. According to this, more able workers shocks motivate workers (low skilled ﬁrst) to leave their job will sort into occupations with higher non-routine, cognitive to seek higher wage opportunities. Likewise the search and task shares, while less able workers switch to high routine, matching literature (c.f. Pissarides 2000 for an overview) non-cognitive jobs. Therefore, only the more able workers predicts a positive relationship between job mobility and will experience a rise in wages upon a job switch. outside wages, as workers are rational and only move if in- Yet another interpretation for the duality of wage out- centivized. For the German labour market, Fitzenberger and comes upon occupational changes is presented by Gath- Spitz-Oener (2004) ﬁnd an overall positive relationship be- mann and Schönberg (2010) and also Geel and Backes- tween occupational switches and wages, thereby supporting Gellner (2011). They attribute the probability of a wage that occupational mobility mainly serves as a career seeking gain after a switch to the proportion of speciﬁcity of the device. acquired skills in the former occupation. Geel and Backes- However, there is also always a non-negligible share of Gellner (2011) show that the higher the speciﬁcity of skills, job switchers who have experienced downward mobility (cf. the lower occupational mobility. In addition, Gathmann Gibbons and Katz 1991). Whereas voluntary quits are most and Schönberg (2010) also show that occupational mobil- often associated with separations to higher paying jobs, in- ity mostly entails switches to related ﬁelds, where skills are K Modelling reallocation processes in long-term labour market projections 71 best transferable. Apart from the share of speciﬁc human tute for Employment Research (IAB) in collaboration with capital needed in an occupation, Damelang et al. (2015) the Fraunhofer Institute for Applied Information Technol- indicate that also to the degree of standardisation and occu- ogy (FIT) and the Institute of Economic Structural Re- pational closure is important. A higher degree of regulation search (GWS). As this paper focuses on possible reallo- (meaning the existence of occupation speciﬁc VET certiﬁ- cation mechanisms of labour demand and supply to over- cates and study programs) reduces the propensity of leaving come long-term mismatches at the occupational level, we the occupation. will only brieﬂy touch on the derivation of labour demand Additionally, there are of course also other factors driv- and supply in the QuBe projections and describe the im- ing job mobility aside from monetary incentives. Cotton and plemented reallocation mechanism more thoroughly. The Tuttle (1986), Shaw et al. (1998), Pollmann-Schult (2006), reader is referred to Maier et al. (2014, 2015) for a detailed Böckermann and Ilmakunnas (2009), Cottini et al. (2011) description of the model. Note that the working volume is all emphasize the importance of physical and psychological central to the demand side model and results are also avail- hygiene, as well as, a good work life balance for retention of able in aggregate hours of work. However, for simplicity in employees. Furthermore, on more regional level, regional this paper we only focus on results evaluated in the number mobility within an occupation has to be considered as an of persons involved. alternative to occupational mobility (Reichelt and Abraham The underlying model projects a development path (the 2015). baseline scenario) of the German economy into the future Furthermore, note that other mechanisms that do not con- given that the currently observable behavioural patterns and cern occupational mobility may also be used in projection trends in the goods, labour and education market will con- models. Ehing and Moog (2013) point out that the size of tinue on their develop path until 2030. As such, it does the future workforce hinges on assumptions about future not necessarily represent the most likely development, but labour force participation. Zika et al. (2012) suggest that can be understood as an outlook on the possible structure the amount of hours a person wishes to work signiﬁcantly of the future labour market when every market participant impact labour supply, especially in occupations with large keeps on her current path of motion. Using this approach shares of part-time workers. This suggests that one could enables a straight forward interpretation of the results and also implement a mechanism, which assumes workers to makes them easily comparable to outcomes of alternative react to changes in the labour market by altering their par- scenarios. In this spirit, modes of behaviour, which cannot ticipation or their working volume. Also migration ﬂows be empirically veriﬁed, are considered infeasible for the re- could dynamically adjust to the labour market situation in sulting baseline scenario. Thus, for example technological a projection model. However, such mechanisms have not progress is only captured by a constant trend and not as- been implemented in any projection model so far. In the sumed to accelerate until 2030. We do, however, implement QuBe model all of these measures are assumed to be stable future changes which have been enacted by legislation and or to follow a trend in their development. have a relevant effect on the outcome during the projection To sum up, in theory wage impulses should create an period. As an example, the baseline scenario takes the new incentive to switch occupations. However, not all occupa- German pension age of 67 into account. tional switches are found to be associated with an increase Fig. 1 gives a highly simpliﬁed overlook of the QuBe in wages. Therefore, in the aggregate the effect of wages on model. Two concurrent processes essentially determine occupational mobility may be mediated by downward mo- labour market outcomes: The evolution of labour supply bility of a part of the occupation switchers. Indications that driven by demographic change (left box) and the evolution the possibility of downward movements is associated with of labour demand, which is driven by economic structural the nature of the task or the prior income level, suggest that change (right box). Both labour supply and demand devel- wage effects should differ by occupations. In addition, other opments are projected until 2030. Essential to the model factors concerning the perceived attractiveness of the occu- is the distinction between the training occupation, which pation seem to have an important impact of occupational workers are associated with on the supply side, and exer- mobility. cised occupation, which workers relate to on the demand side of the labour market. On the supply side, we project the numbers of new labour 3 The BIBB-IAB qualiﬁcation and occupational supply, those leaving the labour market, and ultimately the ﬁeld projections total supply given their sex, age, qualiﬁcation level, and training occupation. For this purpose, the Fraunhofer FIT In this section, we will describe the underlying model. The developed a cohort component model (c.f. Whelpton 1936; QuBe model is a joint project of the Federal Institute for Blien et al. 1990; more speciﬁcally for QuBe see also Kali- Vocational Education and Training (BIBB) and the Insti- nowski and Quinke 2010), which subdivides the popula- K 72 T. Maier et al. Fig. 1 The QuBe model (Source: QuBe projections; rd 3 wave) tion according to sex, age, and qualiﬁcation characteristics not taken into account. While the short-term may be con- and extrapolates the in- and outﬂows of these subgroups cerned with, for example, dealing with the consequences into the future (BIBB-FIT model). The movements between of the euro crisis, structural change is the essential deter- groups summarize ageing given births and deaths, migra- minant of labour demand in the long-term. In pursuance of tion, and qualiﬁcation attainment behaviour. The latter is accurately reﬂecting structural change, QuBe relies on the simulated with a nested transition model of the German QINFORGE model developed by the GWS – a further de- education system. Here, pupils are allocated and transition- velopment of the IAB-INFORGE model (Meyer et al. 2007; ing between high school tracks, entering the vocational ed- Schnur and Zika 2009; Maier et al. 2015). QINFORGE is ucation system, switching between higher education and an econometric input-output model for Germany, which is vocational training programs and, ﬁnally, according to the overall completion rates of the different programs ﬁnishing 2 Vacancies are not taken into consideration in the QuBe long-term by obtaining a credential assigning them to a qualiﬁcation projections for four reasons:. Micro-macro problem: At the micro-economic level, the non-ﬁlling level and according to the prevailing empiric rates of oc- of a vacancy leads to a loss if it causes the company concerned to refuse cupation attainment a training occupation, which they can orders and, thus, to restrict or not to expand production capacity. This use in the labour market to earn wage proﬁts. Of course, does not, however, necessarily mean that there is a corresponding loss infeasible transitions which cannot be identiﬁed in the data in production for the economy as a whole, i. e. at the macro-economic level. Indeed, it may instead lead to the acceptance of the order by are not considered. Note further that people in or without another domestic company, which instead expands its production ca- any vocational education do not have a training occupation pacity, offsetting the potential loss in demand. by deﬁnition and can, therefore, only be associated with an Methodology: Without further background knowledge, no expansion exercised occupation if they are economically active. The demand can be deduced solely from an increase in vacancies, since the number of vacancies cannot be differentiated according to replacement number of economically active persons for each subgroup and expansion demand. is calculated using group speciﬁc participation rates, which Long-term observation: From an economic point of view, vacancies are forecasted with a logistic trend model. only become a problem – if at all – if they cannot be ﬁlled. Even if On the demand side, we calculate the total number of we do not impute complete information or rational agents, problems with an unﬁlled vacancy should vanish with time as a result of the persons needed to manufacture and provide the total num- reallocation process. Therefore, we safely that the number of vacancies ber of goods and services produced in Germany given their always returns to its frictional level in the long term. qualiﬁcation and exercised occupation for each economic Data quality: Reported vacancies statistics by the Federal Employ- sector. We refer to this as realised demand; vacancies are ment Agency (BA) also contain vacancies that do not have to be ﬁlled necessarily. The reasons for this may be multifarious: neglect of report- ing a successful ﬁlling by the company or duplicate reports. Although this problem does not arise with data of the Job Vacancy Survey con- ducted by the IAB, the data here is not available to a sufﬁcient depth of occupational disaggregation. K Modelling reallocation processes in long-term labour market projections 73 ld deeply disaggregated by economic sectors and commod- w = ˛ + ˛ W + ˛ (1) o 1 2 3 ls ity groups. To describe this model in a very simpliﬁed way, let state, employers, and private households invest In a further step, the industry- and occupation-speciﬁc and consume, thereby generating demand. On top, there wage (w ) is modelled. Here, note that the QuBe model as- o;i is a demand for German products from abroad. Also, in- sumes an underlying productivity-based wage policy. Thus, ternational trade poses price pressures on exports and im- industry level wage differences within occupations are ex- ports, which affect price levels for consumption but also plained by differences in labour productivity. Thus, production goods in Germany. This affects the demand for imported goods and also raises unit costs for German prod- w = ˇ + ˇ w + ˇ lpp ; (2) o;i 1 2 o 3 i ucts. Given the individual input-output interdependencies of the economic sectors, the production level is raised or low- where lpp denotes the industry speciﬁc productivity of ered accordingly. Production results in value creation and labour. Again, a constant is included to account for any employment, leading again to a reaction of consumption time invariant determinants of the level of industry- and and investments. In an iterative process these described in- occupation-speciﬁc wages. terdependencies between the different economic actors de- After modelling the wage dependency on labour scarcity, termine the ﬁnal growth path of Germany and the level of the industry and occupation speciﬁc wage is integrated into employment per economic sector, which, according to the the projection of labour demand. Demand for labour by structure of each sector, translates to a demand of labour occupation and industry is explained by the relative appli- for each exercised occupation. cation of the occupation in the economic sector as given Having derived both labour demand and supply, we con- by its contribution to total industry volume of work, i. e. tinue now with a more detailed description of the reallo- occupation- and industry-speciﬁc volume of work relative cation mechanism, which connects both sides (see Fig. 1). to total industry volume of work. The industry-speciﬁc vol- Sect. 3.1 will be concerned with the wage adjustment mech- ume of work is driven by the output level and constraint anism of employers, while Sect. 3.2 will outline the oc- by industry-speciﬁc wage costs. Also, due to technological cupational ﬂexibility adjustment mechanism of workers. progress it is explained by a decreasing time trend indicat- Together both mechanisms form the reallocation process ing the growing efﬁciency of labour inputs. The connection imbedded in the QuBe model. However, we wish to point between volume of work and labour scarcity is modelled out that such a reallocation mechanism could easily be by Eq. 3. transferred to other projection models. vow w o;i o;i = + + t (3) 1 2 3 vow w i i 3.1 Modelling wage adjustment due to skill shortages The equation states that the relative differences in work This section describes the labour demand adjustment mech- inputs between occupations in the same industry is ex- anism through the wage channel with respect to labour mar- plained by a time trend (t) and the relative wage difference o;i ket tightness. Note that the occupation dimension to a very ( ). The latter depends on the occupation speciﬁc labour high extent already captures the informational input of qual- scarcity (cf. Eq. 1). Thus, relatively scarce labour will be iﬁcation. relatively pricy such that its application in the production The starting point is the occupation speciﬁc wage, which process measured by its volume of work is lowered. Given is a function of the total average wage in the economy (W ), that the amount of annual hours worked by one labourer and a scarcity term. The latter is given by the ratio of labour in this industry and occupation does not change, there will demand (ld ) and supply (ls ) in the occupation and op- be a decrease in labour demand in this occupation in this o o erationalizes the overall tightness within the occupational industry. Note that an adverse shock to scarcity causes a per- submarket. W itself is a function of aggregate per capita turbation, since the resulting change in labour demand will labour productivity, overall ﬂuctuation in prices and an ag- in turn alter the scarcity measure again, which moderates gregate term of the labour market tightness for the entire wages and labour demand. Such a perturbation also affects economy. Additionally, a constant is included, which cap- other industry wages through a change in aggregate income. tures all occupation-speciﬁc time invariant factors, which This modiﬁes consumer demand, which is the main driver also determine occupation wages. This captures, for exam- for production in a lot of industries. An increased produc- ple, the extent to which employers could overcome labour tion level induces a raise in labour demand, which again shortages by raising employee productivity by innovative starts off the process of wage adjustments in the affected technologies or further training within a certain occupation industries. (cf. Sect. 2.1). K 74 T. Maier et al. 3.2 Modelling occupational ﬂexibility due to wage 4.1 Data and classiﬁcations adjustments For the QuBe model, data from a number of sources This section outlines the reallocation process of labour sup- was merged to generate a unique data set, which outlines ply on the occupational level through the wage channel. The a deeply disaggregated picture of the German economy and basic idea is that within the model occupational switches are the labour market. For structural information, we rely on accounted for, i. e. it is not assumed that a person who has data of the years 1996 to 2011 retrieved from the German been trained in a certain occupation automatically is part of Microcensus (Labour Force Survey), which is a yearly sam- this occupation-speciﬁc labour supply. Therefore, the start- ple survey of roughly 1% of the German households. It is ing point of modelling this mechanism is the distribution the main source of information for the population structure of the skilled labour force by training occupation over all with regard to age, sex, qualiﬁcation level, employment sta- exercised occupations. Persons, for which the training and tus and training occupation (Maier and Helmrich 2012). It the exercised occupation are identical, are called stayers, also provides data on the distribution of gainfully employed henceforth. The share of stayers in the training occupation, persons over industries and exercised occupations for the to, is denoted by st aye r . years 2005 to 2011 and can, therefore, also be used to anal- to This stayer share is assumed to be time variant and reacts yse occupational switches. Furthermore, it contains data on to impulses of the economic environment. In the model, self-employed and civil servants. No other survey delivers these impulses are captured by outside wage opportuni- a more complete picture for all these characteristics. ties given by a training occupation speciﬁc reference wage On the demand side, information on consumption, prices, ref (w ), which is the weighted average of the wages of all and production for the years 1991 to 2011 is retrieved to (inside and outside) work opportunities, which are feasible from the National Accounts of the Federal Statistical Of- (considering the distribution over exercised occupations) ﬁce (FSO, henceforth). Especially, the input-output-tables for a certain training occupation. The share of stayers is enable a modelling of the interindustry dependencies within determined by equation the production process. For the wage development, we retrieve daily wages for to st aye r = ı + ı (4) full-time employees of the years 1993 to 2011 from the to 1 2 ref to IAB Employment History Data (EHD), which records all where w denotes the wage in the training occupation, to. employment relationships subject to social security contri- to The equation states that whenever a certain training occupa- butions in Germany and captures information about work- tion experiences an increase in wages while the wage level ing days per person and wage totals by economic indus- remains constant in all other reference occupations, it will try, occupation exercised and qualiﬁcation level. By relying become relatively more proﬁtable to stay in the training oc- on this data set, note that we misrepresent wages of civil cupation, thus, causing a rise in the share of stayers. The servants, self-employed and helping family members. Also, extent to which the intent to stay in the training occupa- wages of top income earners are underestimated due to legal tion reacts to outside wage pressures is determined by ı , censorship in the upper income range. However, employees which is the training occupation-speciﬁc wage elasticity of subject to social insurance contributions represent the ma- the propensity to stay. Again, a wage rise triggers a pertur- jority of the work force (about 89% in 2015) and there is bation, where the aggregate effects on labour supply cause no larger and more detailed dataset on gross wages avail- a re-evaluation of wages and labour demand, which, in turn, able in Germany. We, therefore, use the wage development causes preceding adjustments of the supply side and so on. of the EHD as indicator for the general occupation and in- dustry speciﬁc wage development. Note also, that with the underlying data the new legislation on minimum wages is 4 Operationalization and estimation of the QuBe not yet accounted for. model Furthermore, we use the 12th Coordinated Population Forecast of the Federal Statistical Ofﬁce ‘Version 1–W2: In the following section, we brieﬂy present the data used Upper limit of the “medium” population’ until 2060 to to estimate the QuBe model and point out some indication quantify the population by age and sex in the future. To of the explanatory power of scarcity for labour demand and wages for labour supply, respectively, before we fur- A preliminary assessment of the minimum wage policy based on the th QuBe model was presented on the 11 International Conference Chal- ther highlight the magnitude of their impact by sensitivity lenges of Europe in 2015. The results suggest a negative overall impact analyses in the subsequent section. on the economy. Service-oriented industries and professions with low to medium-skilled qualiﬁcations are likely to be exposed the most. See also URL: https://www.efst.hr/eitconf/index.php?p=proceedings. K Modelling reallocation processes in long-term labour market projections 75 rd be able to account for the current developments in the pop- est available Microcensus when the 3 wave of the QuBe ulation in both absolute terms and in terms of their changed project was computed. Secondly, it was also the last Mi- age structure, Version 1-W2 was adapted to the new results crocensus, which used the KldB92 to classify occupations. of the Census 2011. Note that Version 1-W2 is meant to re- Thereafter a harmonization of past data to the 2010 Classi- ﬂect an upper limit of the population, however, understates ﬁcation of Occupations is needed. the current net migration inﬂows of, in particular, political and religious refugees. Accounting for this is likely to im- 4.2 Estimation of the QuBe model pact the projection outcomes. As an example, the demand for teachers may be increased considering the high share In this section, we brieﬂy outline how the before mentioned of young migrants. Therefore, the QuBe projection results, equations of the reallocation mechanisms were estimated. as well, are outdated in this sense. This illustrates how the Using data from 1993 to 2011 on daily wages of full- plausibility of long-term projections strongly hinges on cur- time employees, working volume and labour productivity, rent beliefs of future developments. However, to show the Eqs. 1 to 3 were estimated adding an error term to the right effects of different modelling assumptions concerning the hand side, where the subscripts o and i are captured by the adjustment process on the projection results it can also be 54 OF and the 63 aggregated economic sectors, respectively. helpful to isolate effects from such factors. We, therefore, The t-test for the parameters of Eq. 1 indicate (at a signif- think that our results can be used to visualize the impact of icance level of 5%) that the measure of labor scarcity is the modelling of the reallocation process, even though the a good, necessary and observable predictor for wage level recent migration behaviour is not taken into account. differences between occupations. Especially for ‘occupa- For the calculation of new labour supply by qualiﬁca- tions concerning the production of chemicals and plastics tion level and formal vocational qualiﬁcation, the forecasts wages’ largely, signiﬁcantly depend on labour market tight- of the Conference of Ministers of Education and Cultural ness. However, in 8 of the 54 OF, the effect of scarcity is Affairs of the Länder in the Federal Republic of Germany found to be insigniﬁcant. An example is the ‘public admin- of pupils and graduates from German high schools and uni- istration occupations’. An explanation could be the lack of versity entrants until 2025 are used as a benchmark for the variation in the scarcity variable in these OF. future development in schools and higher education. The Eq. 2 uses the results of Eq. 1 for estimating occupation- retrieved entry, graduation and transition rates for 2025 are speciﬁc wages in each of 63 industrial sectors. A potential held constant thereafter until 2030. of 3402 wages are estimated accordingly. However, not all For both the supply and the demand side the date is ag- occupation and industry combinations exist: taking 2010 gregated using the same classiﬁcation schemes. The Inter- as base year, only 75% of all possible combinations report national Standard Classiﬁcation of Education 1997 is used employment. The corresponding regressions are estimated to differentiate between qualiﬁcation or skill levels. For using ordinary least squares. The estimated parameters are the occupation dimension, the 369 occupational categories evaluated against the R (greater than 0.90), Durbin-Wat- (3-digit code) of the 1992 Classiﬁcation of Occupations son test statistic (between –1 and 1), and the p-value (be- (KldB92) are aggregated according to the 54 occupational low or at 0.05). In total, it was possible to identify wage ﬁelds (OF, henceforth) of Tiemann et al. (2008). Using the responsiveness in 1.513 occupation-speciﬁc industry wages OF to distinguish between occupations prevents artefacts which means that roughly 30 thousand employees are wage- in the modelling of occupation switches, which particularly sensitive in an econometric sense. Nonetheless, there exist occur in the manufacturing sectors because the KldB92 is some cases for which no conclusions about the existence of very detailed here. For an easier visualisation, we report an industry-speciﬁc penalty or mark-up can be made, be- our results for 20 main occupational ﬁelds (MOF, hence- cause either the coefﬁcient of the industry-speciﬁc labour forth) – an aggregated version of the OF (see Table 5 in productivity is insigniﬁcant or the regression is subject to the appendix). Economic sectors are classiﬁed using the autocorrelation. In these 28% of the cases a default option aggregation to 63 industries of the National Classiﬁcation is used, using the OF wage to update the industry speciﬁc of Economic Activities of 2008 (Table 6 in the appendix). OF wages. A similar approach is used for the estimation of To harmonise the supply and demand side data, the num- Eq. 3, where in cases of autocorrelation or insigniﬁcance of ber of persons in active employment as retrieved from the the wage relation by default the relative inputs of occupa- Microcensus is re-extrapolated to match the total number tions is kept constant. Therefore, not in all cases changes in as recorded in the National Accounts, while retaining the the labour supply transmit a change in wages and likewise structure of the population by age, sex, educational level not all wage changes induce a change in the occupational and formal vocational qualiﬁcation from the Microcensus. structure of the industry. For the estimation of Eq. 4, ﬁrstly, Throughout, 2011 is the base year of the QuBe projection. the distribution of formally trained workers by 54 training The reason is that ﬁrstly, the Microcensus 2011 was the lat- OF over the exercised OF is calculated for each age, sex K 76 T. Maier et al. Table 1 Occupational ﬂexibility matrix from formal vocational qualiﬁcation to occupation exercised in 2011 for 20 MOF MOF formal vocational Switches to MOF exercised (in %) qualiﬁcation 1234 5 6 789 10 11 12 13 14 15 16 17 18 19 20 Total 1 Raw material pro- 51.1 3.1 1.6 3.5 0.6 2.9 3.2 2.5 10.9 1.2 2.1 4.2 4.9 0.9 1.2 2.6 0.9 1.2 0.8 0.6 100 cessing occupation 2 Auxiliary workers, 0.0 66.3 6.6 6.9 0.0 0.0 0.0 1.4 2.6 1.7 0.0 1.7 2.6 0.0 2.9 4.6 1.3 0.0 0.0 1.5 100 janitors 3 Metal production 1.4 5.8 36.6 3.6 1.3 9.8 1.3 2.6 10.2 2.2 1.1 2.0 3.8 3.4 8.5 3.5 1.4 0.7 0.4 0.5 100 and processing, in- stallation, electrical occupations 4 Construction, wood- 2.1 5.8 3.2 46.9 1.9 5.5 1.7 2.0 14.0 2.3 1.1 2.8 2.7 1.0 2.1 2.3 0.9 0.8 0.5 0.5 100 working, plastic manuf. occupations 5 Other processing, 1.8 4.1 4.3 3.4 25.3 7.6 5.4 3.7 16.6 2.1 3.3 7.2 4.6 0.9 2.9 2.6 1.2 1.6 0.7 0.6 100 producing and main- taining occupations 6 Machinery and 1.2 3.5 7.4 2.7 3.1 41.3 2.0 2.5 10.8 2.1 1.6 3.5 3.9 1.8 5.2 2.7 2.8 0.9 0.5 0.5 100 equipment steering/ maintenance occup 7 Commodity trade in 1.4 1.9 1.1 0.2 0.2 1.1 49.5 4.3 5.9 0.5 6.0 12.1 7.9 0.2 0.2 1.6 0.8 3.5 1.3 0.3 100 retail 8 Commodity trade 0.6 1.2 0.6 0.3 0.3 0.9 13.6 34.6 5.4 1.1 3.3 3.5 21.1 1.3 0.6 6.4 2.1 1.5 1.0 0.7 100 merchandise 9 Transport, ware- 1.3 2.7 1.7 2.2 0.9 2.3 2.4 3.0 58.0 2.1 2.0 3.5 10.3 1.2 1.2 1.9 1.1 1.0 0.7 0.6 100 house operatives, packers 10 Personal protection, 0.3 1.0 0.2 1.0 0.1 0.6 0.6 1.4 3.0 80.0 0.8 0.9 4.0 0.6 0.7 2.5 0.5 0.8 0.1 0.9 100 guards and secutrity occupations 11 Hotel, restaurant 3.2 2.1 0.9 1.0 0.6 2.3 6.2 2.9 7.4 1.3 47.5 8.5 6.6 0.5 0.7 3.0 1.2 2.4 1.2 0.8 100 occupation, house- keeping 12 Cleaning, disposal 1.8 3.0 1.5 1.4 0.6 3.3 2.9 1.2 7.2 1.3 5.3 60.9 2.0 0.4 0.5 2.7 0.4 1.9 1.5 0.4 100 occupations 13 Ofﬁce and com- 0.5 0.7 0.4 0.2 0.2 0.5 3.1 5.5 2.6 1.7 1.9 2.1 67.5 1.6 0.7 5.9 1.9 1.4 1.1 0.6 100 mercial services occupations 14 IT and natural sci- 0.7 0.5 1.0 0.5 0.2 0.6 0.8 3.0 1.7 0.8 0.9 0.8 7.3 52.2 4.2 13.2 5.2 0.9 0.5 4.9 100 ence 15 Technical occupa- 0.9 2.0 7.6 1.6 5.9 5.0 2.5 3.6 5.0 1.4 1.7 2.5 8.2 6.2 33.9 6.8 1.9 1.3 0.7 1.5 100 tions Modelling reallocation processes in long-term labour market projections 77 and qualiﬁcation group for the years 2005 to 2011 using Microcensus data. Table 1 shows the aggregate distribu- tion, the so-called ﬂexibility matrix, for the year 2011 for the summarized 20 MOF, where the dark cells highlight the percentage of stayers. Overall, we can see that some groups of persons as distinguished by training OF are more concentrated on fewer exercised OF than others. MOF 20 ‘teaching occupations’ is a classic example of high concen- tration. Next, the elasticity ı of Eq. 4 is retrieved, estimating a model of the log share of stayers on the log wage to reference wage ratio, a constant and an error term. We es- timate this model using the aggregated ﬂexibility matrices over all age, sex and qualiﬁcation groups for the 54 OF cross-sections and the years 2005 to 2011. For more robust results we pool OF of similar qualiﬁcation proﬁles and his- toric wage responsiveness together to estimate this model as four separate ﬁxed-effects panel models. Therefore, in each panel all persons associated with a certain training OF react in the same manner to wages in the model. How- ever, the differences in occupational mobility according to age, sex and qualiﬁcation are accounted for by using the different ﬂexibility matrices for each group in the projec- tion. Panel 1 comprises different OF who have shown high wage responsiveness in the past and consist of high shares of highly educated and very low shares of non-formally qualiﬁed workers. Panel 2 includes highly wage responsive OF with a workforce highly centred in the medium but also in the low qualiﬁcation levels. Panel 3 consists of low wage responsive OF with a similar qualiﬁcation make-up as panel 2. Finally, panel 4 contains miscellaneous OF with historically very low wage responsiveness. Table 2 displays the results of the separate panel regres- sions. Note that we only ﬁnd an elasticity for 36 of the 54 cases. The remaining cases, as for example ‘health-care occupations not requiring a medical practice license’, for which no signiﬁcant elasticity can be found, do not react to wages in the model. In addition, people without any formal qualiﬁcation are assumed to distribute over exercised OF, in which they comprised at least 3% of the workforce in 2011, according to labour demand, while the distribution over exercised OF of those in education are held constant in the projection. It is also likely that the structure of the wage data plays a role in this case. The wage data of gainfully employed persons and the legal censorship in the upper income range probably do not represent an ideal measurement, particularly with regard to the OF of ‘managing directors, auditors, management consultants’ and ‘legal occupations’. In the case of ‘health-care occupations not requiring a medical practice license’, for example, which also show a higher proportion of self- employed persons and a higher income, no positive elasticities can be demonstrated. Nevertheless, because of the absence of a more exact database, it seems appropriate to use the elasticities as given in Table 4 for the baseline projection. Table 1 Occupational ﬂexibility matrix from formal vocational qualiﬁcation to occupation exercised in 2011 for 20 MOF (Continued) MOF formal vocational Switches to MOF exercised (in %) qualiﬁcation 12 34 5 6789 10 11 12 13 14 15 16 17 18 19 20 Total 16 Legal, management 0.4 0.3 0.2 0.2 0.1 0.2 1.3 6.6 1.1 0.8 1.0 0.5 25.1 3.6 0.7 49.3 4.6 0.7 1.0 2.4 100 and business science 17 Media, arts and 0.4 0.6 0.3 0.5 0.5 0.7 2.3 4.0 1.7 0.7 1.8 1.3 9.9 6.0 1.3 7.8 43.9 1.7 2.8 11.9 100 social science 18 Health occupations 0.4 0.7 0.3 0.1 1.3 0.6 3.3 1.6 1.8 0.5 2.0 3.4 6.5 0.5 0.3 1.5 1.0 71.2 2.2 0.9 100 19 Social occupations 0.4 0.5 0.2 0.1 0.1 0.3 1.6 1.0 1.2 0.4 1.6 2.7 5.0 0.5 0.1 2.4 1.7 3.9 66.4 10.0 100 Teaching occupa- 0.2 0.2 0.2 0.2 0.1 0.3 0.8 1.3 1.1 0.2 1.2 1.8 3.9 0.9 0.2 1.5 2.5 1.7 3.4 78.4 100 tions Without formal 3.0 7.8 3.7 4.3 1.7 5.6 7.0 3.1 14.8 1.9 9.9 17.5 7.3 1.2 1.0 2.0 2.2 3.5 1.8 0.8 100 vocational qualiﬁca- tion Still in training 2.0 0.8 6.8 3.9 3.3 4.2 7.2 7.0 6.0 1.5 8.6 1.9 15.6 3.7 2.2 1.7 5.2 10.8 4.2 3.5 100 Source: German Mikrocensus 2011, own calculations (BIBB) 78 T. Maier et al. Table 2 Wage elasticitiy of stayers ı by OF (2005–2011) OF ı Panel 1: 2.2 21 Engineers | 22 Chemists, physicists, scientists | 31 Advertising specialists | 36 Administrative occupations in the public industry | 51 Journalists, librarians, translators, related academic research occupations | 46 Designers, photographers, advertising creators | 24 Technical draughtsmen/draughtswomen, related occupations Panel 2: 2.59 16 Cooks | 34 Packers, warehouse operatives, transport processors | 40 Auxiliary ofﬁce occupations, telephone operators | 52 Body care occupations Panel 3: 1.27 1 Agriculture, husbandry, forestry, horticulture | 2 Miners and mineral extraction workers | 5 Paper manufacture, paper processing, printing | 9 Vehicle and aircraft construction, maintenance occupations | 10 Precision engineering and related occupations | 14 Bakers, pastry cooks, production of confectionary goods | 15 Butchers | 18 Construction occupations, wood and plastics manufacture and processing occupations | 41 Personal protection, guards| 49 Social occupations | 54 Cleaning and disposal occupations Panel 4: 0.57 6 Metal production and processing | 7 Metal, plant, and sheet metal construction, installation, ﬁtters | 13 Textile processing, leather man- ufacture | 17 Production of beverages, foods and tobacco, other nutrition occupations | 23 Technicians | 25 Surveying and mapping | 26 Specialist skilled technicians | 27 Sales occupations (retail) | 30 Other commercial occupations (not including wholesale, retail, bank- ing) | 32 Transport occupations | 35 Managing directors, auditors, management consultants | 39 Commercial ofﬁce occupations | 44 Legal occupations | 53 Hotel and restaurant occupations, housekeeping Source: German Mikrocensus and EHD from 2005 until 2011; own calculations Note that the result that workers and employers of dif- changes. In the QuBe model, mainly in favour for keeping ferent training occupations and different economic sectors, the model simple such that results are more transparent, respectively, do not adjust to changes in the labour market in this, however, is not accounted for. the same magnitude, conforms to the discussion of Sect. 2: The reallocation process is also subject to inﬂuences other 5 Scenario comparisons than wages. These are (only) implicitly contained in the QuBe model. However, the comparison of these wage elasticities to In this section, we will display the magnitude of effect results of other studies is limited. The reasons are that (a) of the previously described reallocation mechanism of the these elasticities do not resemble causal effects, but also QuBe model on the projection outcomes and the practi- capture other effects which relate to wages and mobility; cal recommendations based on them. For this purpose, we and (b) because they are based on the relation of the oc- estimate labour demand and supply for various scenarios cupation speciﬁc reference wage with the stayer rate (see concerning a different occupational ﬂexibility behaviour or Eq. 4). Because the reference wage contains also the own wage setting assumptions. Firstly, in Sect. 5.1 we demon- wage of each occupation proportional to the historic ﬂex- strate the overall effect on the projection results from con- ibility, this relation is higher than only looking at outside sidering versus not considering a reallocation process. Sec- wages. Therefore, these elasticities are relatively high. ondly, in Sect. 5.2 we show, which effect can be attributed Further, these wage elasticities of the stayer rate are kept to the dynamics of worker adjustments with respect to constant over the projection period. Departing from this wages. After this, we continue with scenario comparisons assumption would potentially also relevantly affect the pro- to highlight the limitations to wage adjustments in resolv- jection outcomes. It is plausible, for example, that tech- ing labour shortages in the QuBe model and by that the nological progress has an impact on the extent to which importance of other determinants for occupational mobility, wages drive mobility decisions. New technologies are sug- which are only implicitly modelled. We show that these lim- gested to lead to either an increase of complexity of tasks to itations have a meaningful impact on the deduction of rec- be performed by workers or a ‘deskilling’ of tasks, where ommended actions to alleviate occupation-speciﬁc labour specialized skills become redundant (cf. Ben-Ner and Ur- shortages. For this purpose, thirdly, in Sect. 5.3 we show tasun 2013). A change in the skill requirements may lead how the economic environment of the employer matters for to a change in mobility behaviour following the reason- the result of different wage setting policies and the feasibil- ing of Geel and Backes-Gellner (2011) and Gathmann and ity of such wage scenarios according the the QuBe model. Schönberg (2010). Different outside opportunities may then Lastly, in Sect. 5.4 we complement the previous result by also translate into a different receptiveness for relative wage deriving the optimal stayer rates for the occupations and K Modelling reallocation processes in long-term labour market projections 79 rd Fig. 2 Skill shortages and surpluses with and without reallocation in 2005–2030. Source: QuBe project 3 wave; own calculations discuss to what extent these stayer rates are achievable by MOF 18 ‘Health occupations’. The technicians are frequent the means of wage policies. Note that throughout the fol- movers with a stayer share of only 33.9% (cf. Table 2)and lowing section, we implement the scenario assumptions on are able to ﬁnd work in a lot of different MOF. Also, the the level of the 54 OF. However, for a better visualization supply of skilled technicians is decreasing strongly until the results are always presented on the level of the 20 MOF. 2030 (see the decreasing surplus in the left hand graph over time) due to demographic change and retirement of 5.1 Implementing occupational ﬂexibility the so-called ‘baby boomers’, who are more often trained in a manufacturing or technical occupation than younger To start with, Fig. 2 illustrates the effect of implement- cohorts. ing a reallocation process by comparing the projection re- The health occupations, however, face another problem: sults of the QuBe baseline scenario (right hand side) with Workers in this ﬁeld are to a great extent loyal to their a scenario, in which workers were not allowed to switch occupation as indicated by their stayer rate of 71.2% (cf. and employers could not substitute skilled for unskilled or Table 2). Here as well, not enough workers are being trained workers from different disciplines (left hand side). In the in this ﬁeld (again note left hand graph), while the demands latter scenario, the projection results suggest that vast labour are increasing due to the ageing of the population (Maier shortages are possible in 9 out of the 20 MOF. According and Afentakis 2013). to this, for 8 of these MOF shortages should have actu- Ultimately, the total deﬁcit in the baseline scenario is ally already been visible in 2010. In 2030 the deﬁcit would 0.3 million workers only, thus, revealing the substantial grow to about 4.9 million skilled workers in this scenario. In impact of a reallocation mechanism on the projection re- comparison, taking the reallocation mechanism into account sults. Therefore, not taking the empirically veriﬁable oc- balances the labour market in all but 4 of these occupations; cupational mobility into account at all would exaggerate however, shortages appear until 2030 in 5 additional MOF. possible future shortages. Interestingly, now shortages could become especially im- minent in the MOF 15 ‘Technical occupations’ and the K 80 T. Maier et al. in the MOF 5 ‘Other processing, producing and maintain- ing occupations’ (HHI = 0.12). This MOF contains, for example, the textile processors, which have to switch occu- pations more often as the textile industry in Germany is be- ing downsized. Only persons currently in education (HHI = 0.08) and persons with no vocational training (HHI = 0.09) were more evenly distributed. We found the highest concen- tration in the MOF 10 ‘Personal protection, guards and se- curity occupations’ (HHI = 0.64). Also the MOF 18 ‘Health occupations’ (HHI = 0.52) and the MOF 20 ‘Teaching oc- cupations’ (HHI = 0.62) are highly concentrated. These 3 MOF have also the highest stayer rates. The mean HHI equals 0.32 weighted by the labour force participants in each training MOF. In Fig. 3, we now contrast the difference between con- Fig. 3 Differences in HHI due to structural change (2030–2011) and wage development (‘no wage response’ vs. ‘baseline’). Source: QuBe stant and wage responsive ﬂexibility. On the vertical axis, rd project 3 wave; own calculations we plot the pure time trend of the HHI in the 20 MOF, i. e. the HHI in 2011 compared to 2030 of the ‘no wage 5.2 Implementing ﬂexibility dynamics response’ scenario. On the horizontal axis, we plot the HHI differences in 2030 between the baseline scenario Next, we will further analyse how the wage dynamics of with wage elastic ﬂexibility and that without. Note how occupational mobility as implemented in the baseline sce- shifts along the vertical axis visualize pure structural ef- nario of the QuBe model impact the projection results. For fects, while shifts along the horizontal line show how the this purpose, consider a world, in which workers did not concentration of the workforce on exercised occupations respond to wage changes, even if they occurred in occu- increases or decreases as a result of wage incentives. pations in which they could have very likely also found Fig. 3 illustrates that concentration hardly changes over work and proﬁted from a wage gain. Thus, in such a world time due to changes in the labour force composition. An the probability to stay and switch are time invariant. How- exception is the MOF 10 ‘Personal protection, guards and ever, note that aggregate mobility in the occupations does security occupations’. This MOF interestingly has the high- change over time, as the age and qualiﬁcation composi- est HHI in 2011, which, however, is decreased by almost tion of the workforce changes due to demographic change. –0.05 units due to structural change only. Note that the Therefore, comparing projection results for such a world other outlier of MOF 2 ‘Auxiliary workers, janitors’ is ac- with the QuBe baseline scenario enables us to disentangle tually very small in terms of trained labour supply. The the effect of wage responsiveness from structural effects. wage mechanism of the baseline scenario leads to a higher To visualize the concentration of the workforce, i. e. degree of dispersion over exercised MOF in most training the possibilities to work with a certain formal vocational MOF. Wage responses cause the highest reduction in con- qualiﬁcation in different OF, we calculate the Herﬁndahl- centration in the MOF 12 ‘Cleaning, disposal occupations’ Hirschman-Index (HHI henceforth; cf. Hirschmann 1964) and the MOF 18 ‘Health occupations’. Here, the projected for the 20 MOF. wage growth fails that of alternative occupations in other MOF leading to higher occupational switching and, there- fore, a greater dispersion. Note that the observed effect on HHI = (5) to P the MOF 18 can be purely attributed to a change in disper- o=1 o=1 sion of body care occupations, as doctors and nursing staff where x represents the amount of workers in the exercised do not dynamically respond to wages in the baseline sce- MOF o with the training MOF to for which the HHI is nario (cf. Table 2). Also note that the MOF 12 and MOF 18 evaluated. As there are 20 MOF over which the labour force still have some of the highest stayer shares in 2030. In participants of a training occupation can disperse, HHI 2 contrast, in the MOF 16 ‘Legal, management, and business Œ1=20; 1, where the minimum value of 0.05 indicates an science occupations’ or 19 ‘Social occupations’ the wage even distribution over all exercised MOF and the maximum related increase in concentration level out the dispersion value of 1 indicates perfect concentration on the training due to structural effects, such that these occupations have occupation. almost stable HHIs over time. For the year 2011, the ﬂattest empirical distribution is The resulting labour demand and supply for each sce- observed for persons with a formal vocational qualiﬁcation nario in 2030 can be retrieved from Table 3. It can be K Modelling reallocation processes in long-term labour market projections 81 Table 3 Labour demand and supply in 1000 persons by 20 MOF in 2030 in the baseline and ‘no wage response’ scenario Main Occupational Field Baseline ‘No wage response’ (MOF) Supply Demand Diff Supply Demand Diff 1 Raw material processing occupations 907.6 898.6 8.9 929.8 899 30.8 2 Auxiliary workers, janitors 1088.3 1103.6 –15.3 1073 1103.4 –30.5 3 Metal production and processing, installation, 1603.9 1576.5 27.3 1601.7 1575.9 25.8 electrical occupations 4 Construction, woodworking, plastic manufac- 1442.2 1451.1 –8.9 1466.8 1450.6 16.2 ture and processing occupations 5 Other processing, producing and maintaining 848.4 831.9 16.4 858.2 831.7 26.4 occupations 6 Machinery and equipment steering and main- 1772.2 1675.1 96.9 1890 1674.8 215 tainance occupations 7 Commodity trade in retail 2056.5 2059.7 –3.2 1968.5 2058.1 –89.5 8 Commodity trade merchandise 2154.4 2024.3 130.1 2213.5 2023.8 189.7 9 Transport, warehouse operatives, packers 3165.1 3194.6 –29.5 3038.8 3194.2 –155.2 10 Personal protection, guards and security occu- 677.5 657.6 19.9 681.3 657.6 23.7 pations 11 Hotel, restaurant occupation, housekeeping 2158.3 2176 –17.6 2079.3 2170.5 –91.1 12 Cleaning, disposal occupations 2051.7 2052.8 –1.1 2004.4 2052.6 –48.3 13 Ofﬁce and commercial services occupations 6301.3 5545.5 755.8 6479.6 5548.7 931 14 IT and natural science 2369.6 2214 155.6 2390.5 2213.7 176.7 15 Technical occupations 1188.7 1249.2 –60.6 1152.6 1248.9 –96.4 16 Legal, management and business science 2850.8 2679.1 171.7 2811.6 2678.9 132.6 17 Media, arts and social science 1747.9 1792.9 –44.9 1740 1792.2 –52.1 18 Health occupations 3863.7 4016.6 –152.9 3847.9 4027.3 –179.5 19 Social occupations 1739.5 1662 77.4 1737 1662.6 74.4 20 Teaching occupations 1790.6 1500.3 290.4 1813.8 1501.6 312.2 rd Source: QuBe project 3 wave; own calculations observed that without accounting for wage responsive ﬂex- 5.3 The limitations to wage adjustments ibility behaviour, the total deﬁcit equals about 740,000 per- sons. This is more than twice the deﬁcit of the baseline We now examine the impact of wage policies in greater scenario with dynamics, which amounts to only 340,000 detail and point out the importance of their limitations in persons. Thus, 400,000 workers, which would be unem- the QuBe model for the interpretation of results. Shortages ployed in other surplus occupations in the projection, are are partly projected, due to inferior wage developments in redistributed to the shortage occupations where wages are these occupations. Because outside wage opportunities are rising in the baseline scenario. growing more strongly than in the own training OF, work- However, in the MOF 4 ‘Construction, woodworking, ers – where empirically veriﬁable – more often decide to plastic manufacture and processing occupations’ the labour switch occupations. Employers can take advantage of this market actually gets tighter due to wage dynamics. Here, by raising wages in occupations where labour is scarce. although the wage responsiveness of ﬂexibility is actually However, they are (depending on the industry) constraint not too high, the projected development of the outside wage by price competition with ﬁrms abroad and consumer de- options induces the workforce to switch more often to other mand. This is reﬂected in the QuBe model. To show to what occupations. In this case, the possibility of a future labour extent employers can strategically use wage adjustments in shortage may be understated when dynamic behaviour in this model, we implement further wage increases for short- occupational ﬂexibility is not accounted for. Ultimately, we age occupations (as singled out by the baseline projection can conclude that assumptions about the wage responsive- results). We consider a scenario, where wage growth in the ness of labour mobility are crucial for assessing possible shortage occupations is increased by 10% until 2030 com- future labour market outcomes. pared to the baseline wage development. Note that this rep- resents an increase of a little more than 0.5% every year un- til 2030 on top of the projected wage growth of the baseline scenario. Since this represents a relatively small change, in K 82 T. Maier et al. Table 4 Labour demand and supply in 1000 persons by 20 MOF in 2030 in baseline model and different wage scenarios Main Baseline ‘10%-increase’ ‘20%-increase’ Oc- Supply Demand Diff Supply Demand Diff Supply Demand Diff cu- 1 Raw material processing occu- 907.6 898.6 8.9 906 895.6 10.4 904.5 892.6 11.9 pa- pations tional 2 Auxiliary workers, janitors 1088.3 1103.6 –15.3 1090 1097.8 –7.7 1091.6 1092.4 –0.8 Field (MOF) 3 Metal production and process- 1603.9 1576.5 27.3 1602.1 1564.7 37.4 1600.1 1553.4 46.7 ing, installation, electrical occu- pations 4 Construction, woodworking, 1442.2 1451.1 –8.9 1448.4 1433.8 14.6 1453.9 1417.6 36.3 plastic manufacture and pro- cessing occupations 5 Other processing, producing 848.4 831.9 16.4 843.5 826.5 17.1 839.2 821.3 17.9 and maintaining occupations 6 Machinery and equipment 1772.2 1675.1 96.9 1767.5 1663.3 104.2 1763 1652 110.9 steering and maintainance oc- cupations 7 Commodity trade in retail 2056.5 2059.7 –3.2 2047.1 2029.9 17.2 2037.5 2001.9 35.6 8 Commodity trade merchandise 2154.4 2024.3 130.1 2152.2 2010.2 142 2150.2 1996.6 153.6 9 Transport, warehouse opera- 3165.1 3194.6 –29.5 3170.2 3176.3 –6.2 3174.1 3159.3 14.8 tives, packers 10 Personal protection, guards and 677.5 657.6 19.9 681.5 660.1 21.4 685.3 662.2 23 secutrity occupations 11 Hotel, restaurant occupation, 2158.3 2176 –17.6 2111.9 2094.7 17.2 2071.4 2024.6 46.8 housekeeping 12 Cleaning, disposal occupations 2051.7 2052.8 –1.1 2045.6 2029.1 16.6 2038.9 2007 31.9 13 Ofﬁce and commercial services 6301.3 5545.5 755.8 6296.2 5538.6 757.6 6291.8 5531.5 760.3 occupations 14 IT and natural science 2369.6 2214 155.6 2361.4 2207.1 154.3 2353.4 2200.4 153 15 Technical occupations 1188.7 1249.2 –60.6 1202.6 1243.1 –40.4 1214.1 1237.2 –23.1 16 Legal, management and busi- 2850.8 2679.1 171.7 2849.6 2667.8 181.8 2848.8 2657.1 191.7 ness science 17 Media, arts and social science 1747.9 1792.9 –44.9 1766.1 1789.2 –23.1 1782.7 1785.2 –2.5 18 Health occupations 3863.7 4016.6 –152.9 3899.6 4017 –117.4 3933.5 4016.4 –83 19 Social occupations 1739.5 1662 77.4 1740.3 1673 67.3 1741.9 1683.2 58.7 20 Teaching occupations 1790.6 1500.3 290.4 1796.3 1513.6 282.7 1802.2 1526 276.2 rd Source: QuBe project 3 wave; own calculations a second scenario we increase wage growth in the shortage by even 225,000 persons compared to the baseline scenario. occupations by 20%, i. e. an additional increase of a lit- However, the labour market is balanced in only one addi- tle more than 1% every year until 2030. The results are tional MOF compared to a 10% increase until 2030, namely presented in Table 4. the MOF 9 ‘Transport, warehouse operatives, packers’. We The results show (cf. Table 4) that with a wage increase can see that the balance in these MOF is mainly achieved by of an additional 10% until 2030 for shortage occupations, a reduction in labour demand. Since labour productivity re- labour shortages will be reduced by about 140,000 per- mains unchanged, note that this corresponds to a reduction sons in 2030, so that the total deﬁcit in this scenario equals in production or service provision, respectively. In these 195,000 persons. Shortages could be prevented in 4 of the occupations, outside price pressures are too high, such that 9 shortage MOF of the baseline scenario, namely in the large wage adjustments are infeasible for employers without MOF 4 ‘Construction, woodworking, plastic manufacture reducing their output. Here, it is more realistic that alterna- and processing ocupations, MOF 7 ‘Commodity trade in tive strategies would be used to retain workers or workers retail’, MOF 11 ‘Hotel, restaurant occupation, housekeep- would be hired from abroad to keep the wage level low. ing’, and the MOF 12 ‘Cleaning, disposal occupations’. The other shortage MOF, for which a shortage is pro- Looking at the results, of the 20% increase in wage growth jected until 2030 even after an additional wage increase of for baseline shortage occupations, the total deﬁcit of labour 20%, are the MOF 2 ‘Auxiliary workers, janitors’, MOF 15 supply equals about 115,000 persons, which is a reduction ‘Technical occupations’, MOF 17 ‘Media, arts and social K Modelling reallocation processes in long-term labour market projections 83 of practical advice based on calculations of a projection model. 5.4 The ‘optimal’ ﬂexibility In the following, we examine, what kind of adjustments in the occupational ﬂexibility behaviour would be needed to distribute unemployed workers evenly and to overcome labour shortages in every OF in 2030. Thus, this scenario entails a redistribution of the labour supply from surplus to shortage occupations. Also looking at the results from the previous section, we assess how wages can serve to achieve the resulting differences in the stayer rate according to the assumptions of the baseline scenario. Technically, we apply a RAS procedure. The RAS al- Fig. 4 Needed adjustments of occupational ﬂexibility to achieve rd equal unemployment rates in 2030. Source: QuBe project 3 wave; gorithm (cf. Bacharach 1970; Leontief 1951) is an itera- own calculations tive method of biproportional ﬁtting of matrices, which is used to estimate elements of an unknown matrix based on science occupations’, and MOF 18 ‘Health occupations’. known row and column sums and an initial estimate of the In all of these MOF, demand remains relatively stable, sug- matrix. Transferred to this exercise, the RAS algorithm ﬁts gesting that here price pressures are less dominating, be- the cells of the ﬂexibility matrix of 2030, such that column cause production or service provision cannot simply be re- totals, i. e. labour supply in the exercised OF, are such that duced. We leave the MOF 2 out of the discussion as they in every OF an equal unemployment rate is achieved. In comprise a very small group of people and are not asso- doing so, the algorithm loops over occupations – starting ciated with dynamic behaviour in the QuBe model mainly with that with the highest unemployment rate – and redis- due to data restrictions. The MOF 15 and 17 have com- tributes the difference between the baseline surplus supply parably lower stayer rates of 39% and 43%, respectively, and that needed to achieve the targeted unemployment rate because their labour can be applied in very diverse ﬁelds. to other occupations. The reallocation is proportional to the Here, the deﬁcit is more severe in the MOF 15, mainly be- initial ﬂexibility matrix of the baseline scenario, such that cause workers trained in this ﬁeld react much less to wage workers trained in surplus occupations switch more (have impulses. Most of the occupations in MOF 17 are attached a smaller stayer rate); however, to the same extent into the with a wage elasticity of 2.2 in the baseline scenario, in- same exercised occupations. dicating that career seeking is a major determinant in the Fig. 4 visualizes the change in ﬂexibility again using dif- occupational ﬂexibility behaviour of journalists, designers ferences in the HHI indicating growing or declining concen- etc. In contrast, most of the occupations in the MOF 15 tration in the MOF. Here, the difference in the HHI between only react to wages with an elasticity of 0.57, suggesting 2011 and 2030 in the baseline projection is plotted on the that here other factors as for example better working con- vertical axis against the HHI difference between the 2030 ditions may strongly inﬂuence mobility decisions. workforce of the baseline projection and the scenario using In the MOF 18 it is only the body care occupations, the optimal occupational ﬂexibility matrix on the horizon- which react to wage impulses. Increasing wages cannot tal axis. MOF plotted to the left (right) of the 0 benchmark reduce the shortage of doctors and nursing staff, because on the horizontal axis, indicate a need for a higher (lower) the baseline QuBe projection reﬂects that their fairly high ﬂexibility as compared to the baseline assumptions in order occupational loyalty is not signiﬁcantly driven by wage in- to clear the labour market in 2030. centives. Also, a wage increase in these occupations does Overall, the majority of the MOF actually should be not considerably raise the inﬂow of labour supply into these more ﬂexible in order to correspond optimally to labour occupations from other ﬁelds, which simulates the effect of demand. Especially, persons in the MOF 20 ‘teaching oc- strong working regulations concerning qualifying creden- cupations’ but also the MOF 13 ‘ofﬁce and commercial ser- tials and approbations (see also Pollmann-Schult 2006). vices occupations’, for which vast surpluses are projected Overall, we can conclude here that accounting for the due to demographic change and a rising educational at- limitations to wage setting policies within the projection tainment in these occupations, should more often consider model has signiﬁcant impacts on the feasibility of scenarios switching their occupation in the future. In the MOF 20, aimed at overcoming shortages. This has important conse- the share of stayers would have to be reduced from 79.4 quences for policy consulting and enhances the credibility to 66.6% in 2030. In the MOF 13 a reduction of the share K 84 T. Maier et al. of stayers to 61.6% from its level of 67.2% in the baseline most often is the ultimate aim of long-term labour market projection in 2030 would be needed. Note that this MOF projection. also contains the public administrates, which mainly drive this result, here. They alone would need a reduction of the stayer ratio by more than 12 percentage points. However, 6 Conclusion and discussion in both of these MOF, workers do not react to increases in outside wages in the QuBe model and are very loyal to In this paper, we discuss and illustrate the necessity of im- their training occupations (cf. Table 1 and 2). This poses plementing a dynamic reallocation process of labour sup- a challenge of achieving such a reduction in stayer rates. ply into labour market projections and how the underlying Likely, this could not be accomplished by increasing wages assumptions strongly inﬂuence the plausibility of the pro- in related ﬁelds, as other underlying factors as for example jection results and their interpretation for policy consulting. work place stability or reconciliation of family and work Long-term projections have become very popular for guid- are stronger motivators for high stayer rates in these occu- ance in political decision-making. Therefore, it is essential pations. that the model set-up reﬂects country-speciﬁcs and can draw In contrast, persons trained in the MOF 18 ‘Health oc- a plausible image of the possible future developments. In cupations’ would need to stay in their occupation more Germany, therefore, it is essential that a projection model often. The projected stayer rate of 67.8% in 2030 in the (a) represents the occupational dimension of the German baseline scenario would have to increase to 71.9%. This labour market and (b) reﬂects the extent to which workers complements the results of the previous section: Because skilled in different occupations can be substituted for each switches into these occupations are quite unlikely due to other (Helmrich and Zika 2010). These two aspects are es- work regulations, the needed increase in stayers would only sential for an assessment of possible reallocations of labour be achievable via an even greater occupational loyalty or in- supply in respond to imminent shortages. creased training of new supply. Since outside wages are not The BIBB-IAB qualiﬁcation and occupational ﬁeld pro- signiﬁcantly important to doctors and nursing staff, the re- jections (Maier et al. 2014) is to our knowledge the only sults again stress the impact of other factors, as for example long-term projection model, which explicitly formulate working conditions, on making these occupations more at- such a reallocation process. In this model, the central link tractive for policy recommendations to realize the increase between demand and supply is wages: Employers raise in labour supply. wages in shortage occupations to make work in these ﬁelds Interestingly the shortage MOF 15 ‘Technicians’, would more attractive and workers react to relative changes in their hardly need any ﬂexibility adjustments at all according outside wage opportunities and adjust their intent to stay. to this calculation. Their optimal ﬂexibility would entail The great degree of detail of the model by 63 economic a stayer share of 35.7% in 2030. Therefore, the adjustment sectors and 54 occupational ﬁelds provides a thorough de- from its baseline value of 33.2% would amount to merely scription of the diverse adjustment behaviours of different 2.5 percentage points. Here, the redistribution from surplus groups of market participants. In this way, the projection re- ﬁelds is high enough such that only a small adjustment in sults also implicitly capture reallocation behaviours, which the stayer rate sufﬁces to balance the submarket for techni- are not driven by wage and scarcity, respectively. cians. We ﬁnd that almost 70% of the additional workforce Our results show that not accounting for occupational in this MOF would be recruited from outside (mainly engi- ﬂexibility at all, i. e. not modelling any reallocations in the neers and electrical occupations). Here, wage policies may labour market, would project vast shortages of almost 5 mil- serve to attract workers from related ﬁelds to some extent, lion skilled workers in 9 of the 20 main occupational ﬁelds however the persisting shortages even after large wage in- in 2030. Compared to this, the baseline scenario, which ac- creases (cf. Sect. 5.3) suggest that again working conditions counts for dynamic adjustments on both sides, would only in this ﬁeld may be more promising to target. project a total deﬁcit of about 340,000 workers in 2030. In summary, for an optimal distribution of unemployed However, the reallocation process can be directly linked to workers over the exercised occupations, stayer rates for shortages, which now appear in ‘health occupations’ and many training occupation would have to differ. As already ‘technical occupations’. In both of these main occupational discussed in the previous section, wages are often an infea- ﬁelds inﬂows from other ﬁelds would not sufﬁce to balance sible tool to reach the optimum, here. In the QuBe model, out the outﬂows of skilled workers to other related ﬁelds. alternative determinants for occupational mobility are im- Next, looking at the effect of dynamic adjustments of portant for the interpretation of the results, although they the ﬂexibility behaviour of workers, we compare the base- are only implicitly accounted for. In the end, this is essential line scenario to a scenario, where shares of stayers do not for deriving recommendations for practical actions, which respond to wages. We ﬁnd that dynamics can account for a difference in the deﬁcit of labour supply of about 400,000 K Modelling reallocation processes in long-term labour market projections 85 people in 2030. Shortages in the ‘Construction, woodwork- ups where for example dynamics evolve subject to techno- ing, plastics manufacture and processing occupations’ ac- logical progress are possible and maybe a fruitful ﬁeld for tually become more severe in the projection results after research. However, when advancing model set-ups in these considering a dynamic adjustment of workers. Here, wage ways, the transparency of results has to always be kept in dynamics reﬂect the tension between price and employer mind as well (c.f. Wilson 2001). competition for labour supply. Lastly, in the discussed model, the potential of the of- Furthermore, we illustrate how also the limitations to fered amount of hours by the labour force has been assumed wage dynamic adjustments as captured by the QuBe model to be stable during the projection period (Zika et al. 2012). inﬂuence the interpretation of results and the derived rec- Furthermore, it is assumed that participation rates follow ommendations for practical actions. For this, we look at the an increasing trend and migration inﬂow to Germany is th effect of different wage policies. We compare a 10% and kept constant according to the 12 Coordinated Population a 20% increase of wages until 2030 for shortage occupa- Forecast of the Federal Statistical Ofﬁce. Of course, these tions. We see that in the QuBe model these wage increases measures could in principal also work as dynamic mecha- would be able to balance some occupational submarkets, nisms in long-term labour market projections. In fact, this however, mainly by a reduction of labour demand and, thus, may work better for employers in occupations with strong a lower production or service provision in the economy. For wage setting constraints and workers in occupations with the remaining shortage occupations in these scenarios, we low wage responsiveness. As this has not been done thus discuss how wages as a policy tool simply are not effec- far, in future studies it would be very interesting to assess tive given the QuBe assumptions about wage dynamics of the differences in projection results and policy advice be- occupational mobility. Especially for technicians, doctors, tween obtained from projections using these different mech- and nursing staff other factors related to working conditions anisms. may be more important for political actions. In the case of Open Access This article is distributed under the terms of the health occupations, also working regulations play an im- Creative Commons Attribution 4.0 International License (http:// creativecommons.org/licenses/by/4.0/), which permits unrestricted portant role, which limits the extent to which workers from use, distribution, and reproduction in any medium, provided you give outside can be recruited for this ﬁeld. appropriate credit to the original author(s) and the source, provide a We complement these results further, by calculating the link to the Creative Commons license, and indicate if changes were ‘optimal’ ﬂexibility of the workforce, which would evenly made. distribute unemployed workers over the occupations. We ﬁnd that most of the workforce would have to be more ﬂexible. In contrast, health personnel would need to stay more often within their training occupation. As they do not respond to wages empirically, again working conditions but also increased training of new supply may be more feasible policy implementations. Surprisingly, in the case of techni- cians no large adjustment of mobility behaviour would be needed, because also an increased inﬂow of workers from related ﬁelds would help to balance out deﬁcits of labour supply. In this ﬁeld, the sufﬁcient provision of labour sup- ply may be achieved, both to their own extent, by increasing wages and improving work conditions, but also by provid- ing persons with related educational backgrounds further educational training to enhance speciﬁc needed skills. The results illustrate how for the derivation of plausible policy recommendation also the limitations to reallocations are central to modelling. Based on the QuBe model, how- ever, we can only discuss the relative importance of other driving factors of occupational mobility in light of the re- strictions of the wage dynamics. Therefore, also integrating, for example, working conditions into long-term labour mar- ket projection models may be an intriguing ﬁeld of further studies. Furthermore, throughout our analyses we assume that the response of workers to outside wages in their mo- bility decisions is time invariant. Here, as well different set- K 86 T. Maier et al. Appendix Table 5 Major Occupational MOF OF Fields (MOF) and Occupational 1 Raw material processing 1 Agriculture, husbandry, forestry, horticulture Fields (OF) occupations 2 Miners and mineral extraction workers 2 Auxiliary workers, janitors 20 Auxiliary workers without further speciﬁed task 42 Janitors 3 Metal production and 7 Metal, plant and sheet metal construction, installation, processing, installation, ﬁtters electrical occupations 11 Electrical occupations 4 Construction, woodworking, 18 Construction, woodworking, plastics manufacture and plastic manufacture and processing occupations processing occupations 5 Other processing, producing 3 Stoneworking, construction materials production, ceram- and maintaining occupations ics and glass related occupations 9 Vehicle and aircraft construction, maintenance occupa- tions 10 Precision engineering and related occupations 13 Textile processing, leather manufacture 15 Butchers 6 Machinery and equipment 4 Chemical and plastics occupations steering and maintainance 5 Paper manufacture, paper processing, printing occupations 6 Metal production and processing 8 Industrial mechanics, tool mechanics 12 Weaving occupations, textile manufacturers, textile ﬁnishers 17 Production of beverages, food and tobacco, other nutri- tion occupations 7 Commodity trade in retail 27 Commodity trade in retail 8 Commodity trade 28 Wholesale/retail service occupations merchandise 30 Other commercial occupations (not including wholesale, retail, banking) 9 Transport, warehouse 19 Goods inspectors, dispatch, processing operators operatives, packers 32 Transport and logistics occupations 33 Aviation, shipping occupations 34 Packers, warehouse and transport occupations 10 Personal protection, guards 41 Personal protection, guards and secutrity occupations 43 Security occupations 11 Hotel, restaurant occupation, 14 Bakers, pastry cooks, production of confectionary goods housekeeping 16 Cooks 53 Hotel and restaurant occupations, housekeeping 12 Cleaning, disposal occupa- 54 Cleaning and disposal occupations tions 13 Ofﬁce and commercial 29 Banking and insurance professionals services occupations 36 Administrative occupations in the public sector 37 Finance, accounting, bookkeeping 39 Commercial ofﬁce occupations 40 Auxiliary ofﬁce occupations, telephone operators 14 IT and natural science 21 Engineers 22 Chemists, physicists, scientists 38 Core IT occupations K Modelling reallocation processes in long-term labour market projections 87 Table 5 Major Occupational MOF OF Fields (MOF) and Occupational 15 Technical occupations 23 Technicians Fields (OF) (Continued) 24 Technical draughtsmen/draughtswomen, related occupa- tions 25 Surveying and mapping 26 Specialist skilled technicians 16 Legal, management and 35 Managing directors, auditors, management consultants business science 44 Legal occupations 17 Media, arts and social 31 Advertising specialists science 45 Artists, musicians 46 Designers, photographers, advertising creators 51 Journalists, librarians, translators, related academic research occupations 18 Health occupations 47 Healthcare occupations requiring a medical practice licence 48 Healthcare occupations not requiring a medical practice licence 52 Body care occupations 19 Social occupations 49 Social occupations 20 Teaching occupations 50 Teaching occupations Table 6 Structure of the NACE Rev. 2 Classiﬁcation of Economic Activities used in the Projection Divisions of the economic sectors (collated) 1 Agriculture 2Forestry 3 Fishing 4 Mining, extraction of stones and earth 5 Manufacture of food and drink, tobacco processing 6 Manufacture of textiles, clothing, leather goods and shoes 7 Manufacture of wood, wicker, basket and cork goods (not including furniture) 8 Manufacture of paper, cardboard and of paper and cardboard products 9 Manufacture of printing products, reproduction of sound, picture and data storage media 10 Manufacture of coke and reﬁned petroleum products 11 Manufacture of chemical products 12 Manufacture of pharmaceutical products 13 Manufacture of rubber and plastic products 14 Manufacture of glass products, manufacture of ceramics, processing of stones and earth 15 Metal production and processing 16 Manufacture of metal products 17 Manufacture of computer, electronic and optical products 18 Manufacture of electrical equipment 19 Engineering 20 Manufacture of motor vehicles and motor vehicle components 21 Other vehicle construction 22 Manufacture of furniture and other goods 23 Repair and installation of machines and equipment 24 Energy supply 25 Water supply 26 Sewage, waste disposal, materials recovery 27 Construction sector 28 Motor vehicle trade, maintenance and repair of motor vehicles K 88 T. Maier et al. Table 6 Structure of the NACE Rev. 2 Classiﬁcation of Economic Activities used in the Projection (Continued) Divisions of the economic sectors (collated) 29 Wholesale (not including the motor vehicle trade) 30 Retail (not including retail of motor vehicles) 31 Land transport and transport in pipelines 32 Shipping 33 Aviation 34 Warehousing, other transport service providers 35 Post, courier and express services 36 Hotel and restaurant trade 37 Publishing 38 Audiovisual media and radio 39 Telecommunications 40 IT and information service providers 41 Financial services providers 42 Insurance and pension funding 43 Activities associated with ﬁnancial and insurance services 44 Real estate 45 Legal and tax consultancy, management consultancy 46 Architectural and engineering companies, technical support 47 Research and development 48 Advertising and market research 49 Freelance, scientiﬁc, technical services (not mentioned elsewhere), veterinary medicine 50 Renting of mobile goods 51 Placement and hiring of workers 52 Travel agencies and tour operators 53 Service providers (not mentioned elsewhere) 54 Public administration, defence, social security 55 Education and teaching 56 Healthcare system 57 Residential homes and social services 58 Art and culture, gambling 59 Sport, entertainment and recreation 60 Lobbying, religious associations 61 Repair of computers and used goods 62 Other providers of mainly personal services 63 Housekeeping services Gesellschaft: Problemlagen und betriebliche Reaktionen. IAB References Forschungsbericht 13/2012. (2012) Ben-Ner, A., Urtasun, A.: Computerization and skill bifurcation: the Abraham, M., Damelang, A., Schulz, F.: Wie strukturieren Berufe role of task complexity in creating skill gains and losses. Ind La- Arbeitsmarktprozesse? Eine institutionentheoretische Skizze. bor Relat Rev 66(1), 225–267 (2013) LASER Discussion Papers, vol. 55. Friedrich- Alexander -Uni- Blien, U., Reinberg, A., Tessaring, M.: Die Ermittlung der Übergänge versität, Erlangen-Nürnberg (2011) zwischen Bildung und Beschäftigung. Mitt Arbeitsmarkt Berufs- Acemoglu, D., Autor, D.: Skills, tasks and technologies: Implications forsch 23(4), 181-204 (1990) for employment and earnings. In: Ashenfelter, O., Card, D. (eds.) Böckermann, P., Ilmakunnas, P.: Job disamenities, job satisfaction, quit Handbook of labor economics, vol. 4, pp. 1043–1171. Elsevier, intentions, and actual separations: putting the pieces together. Ind Amsterdam (2011) Relat (Berkeley) 48(1), 73–96 (2009) Allmendinger, J.: Educational systems and labor market outcomes. Eur Bonin, H., Schneider, M., Qunike, H., Arens, T.: Zukunft von Bildung Sociol Rev 5, 231–249 (1989) und Arbeit. Perspektiven von Arbeitskräftebedarf und -angebot Bacharach, M.: Biproportional matrices and input-output change. bis 2020. IZA Research Report, vol. 9. (2007) Cambridge University Press, Cambridge (1970) Brücker, H., Klinger, S., Möller, J., Walwai, U.: Handbuch Arbeits- Bechmann, S., Dahms, V., Tschersich, N., Frei, M., Leber, U., Schwen- markt 2013. Analysen, Daten, Fakten. IAB Bibliothek, vol. 334. gler, B.: Fachkräfte und unbesetzte Stellen in einer alternden (2013) K Modelling reallocation processes in long-term labour market projections 89 Brunow, S., Garloff, A.: Arbeitsmarkt und demograﬁscher Wandel. IAB-Modellrechnungen zu den Entwicklungen in Berufsfeldern Anpassungsprozesse machen dauerhaften Fachkräftemangel un- und Qualiﬁkationen bis 2025, pp. 103–124. Bundesinstitut für wahrscheinlich. IAB Forum 2/2011. (2011) Berufsbildung, Bielefeld (2010) CEDEFOP: Skills supply and demand in Europe. Medium-term fore- Lapointe, M., Charron, M., Claveau, G., Gendron, M., Gomez, E.G., cast up to 2020. Publications Ofﬁce of the European Union, Lux- Grenier, M., Ignaczak, L., Kim, J.-Y., Lamy, R., Pescarus, C., embourg (2009) Tremblay-Côté, N., Vincent, N., Zou, Y.: Looking-ahead: a 10- CEDEFOP: Future skills and demand in Europe. Forecast 2012. Re- year outlook for the Canadian labour market (2008–2017). Hu- search Paper, vol. 26. Publications Ofﬁce of the European Union, man Resources and Skills Development Canada, Gatineau (2008) Luxembourg (2012) Leontief, W.W.: The structure of American economy, 1919–1939. An Cortes, G.M.: Where have the middle-wage workers gone? A study of empirical application of equilibrium analysis. Oxford University polarization using panel data. J Labor Econ 34(1), 63–105 (2016) Press, Oxford (1951) Cottini, E., Kato, T., Westergaard-Nielsen, N.: Adverse workplace con- Lepic, M., Koucky, J.: Employment forecasting in the Czech Repub- ditions, high-involvement work practices and labor turnover: evi- lic. In: Arendt, L., Ulrichs, M. (eds.) Best practices in forecasting dence from Danish linked employer–employee data. Labour Econ labour demand in Europe. IPISS, Warsaw (2012) 18(6), 872–880 (2011) Lockard, C.B., Wolf, M.: Occupational employment projections to Cotton, J.L., Tuttle, J.M.: Employee turnover: a meta-analysis and 2020. Employment outlook 2010–2020. Mon Labor Rev 135(1), review with implications for research. Acad Manag Rev 11(1), 84–108 (2012) 55–70 (1986) Lucas, R.J., Prescott, E.: Equilibrium search and unemployment. Damelang, A., Schulz, F., Vicari, B.: Institutionelle Eigenschaften von J Econ Theory 7, 188–209 (1974) Berufen und ihr Einﬂuss auf beruﬂiche Mobilität in Deutschland. Maier, T., Afentakis, A.: Forecasting supply and demand in nursing Schmollers Jahrb 135, 307–334 (2015) professions: impacts of occupational ﬂexibility and employment Dupuy, A.: Forecasting expansion demand by occupation and educa- structure in Germany. Hum Resour Health 11(1), 24 (2013). tion in the Netherlands. In: Arendt, L., Ulrichs, M. (eds.) Best doi:10.1186/1478-4491-11-24 practices in forecasting labour demand in Europe, pp. 127–147. Maier, T., Helmrich, R.: Creating the initial vocational qualiﬁcation IPISS, Warsaw (2012) from the German microcensus. Paper presented at the ACSPRI Dustman, C., Glitz, A.: How do industries and ﬁrms respond to changes Conferences, RC33 Eighth International Conference on Social in local labor supply? J Labor Econ 33(3), 711–750 (2015) Science Methodology, Sydney. (2012) Dustmann, C., Ludsteck, J., Schönberg, U.: Revisiting the German Maier, T., Zika, G., Mönnig, A., Wolter, M.I., Kalinowski, M., Hänisch, wage structure. Q J Econ 124(2), 843–881 (2009) C., Helmrich, R., Schandock, M., Neuber-Pohl, C., Bott, P., Hum- Ehing, D., Moog, S.: Erwerbspersonen- und Arbeitsvolumenprojektio- mel, M.: Wages and occupational ﬂexibilities as determinants of nen bis ins Jahr 2060. J Labour Mark Res 46(2), 167–182 (2013) the interactive QuBe labour market model. Discussion Papers Fitzenberger, B., Lickleder, S., Zwiener, H.: Mobility across ﬁrms and No. 149. Federal Institute for Vocational Education and Training, occupations among graduates from apprenticeship. IZA Discus- Bonn (2014) sion Paper, vol. 9006. (2015) Maier, T., Mönnig, A., Zika, G.: Labour demand in Germany by in- Fitzenberger, B., Spitz-Oener, A.: Die Anatomie des Berufswechsels: dustrial sector, occupational ﬁeld and qualiﬁcation until 2025 – Eine empirische Bestandsaufnahme auf Basis der BIBB/IAB- model calculations using the IAB/INFORGE model. Econ Syst Daten 1998/1999. In: Franz, W., Ramser, H.J., Stadler, M. (eds.) Res 27(1), 19–42 (2015) Bildung Wirtschaftswissenschaftliches Seminar in Ottobeuren, Mayer, K.U., Carroll, G.R.: Jobs and classes: structural constraints on vol. 33, pp. 29–54. Mohr Siebeck, Tübingen (2004) career mobility. Eur Sociol Rev 3(1), 14–38 (1987) Gajdos, A., Zmurkow-Poteralska, E.: Employment forecasts by oc- McLaughlin, K.J.: A theory of quits and layoffs with efﬁcient turnover. cupational groups in Poland. Polit Spoleczna Themat 2014(2), J Polit Econ 99(1), 1–29 (1991) 14–20 (2014) Meyer, B., Lutz, C., Schnur, P., Zika, G.: National economic policy Gathmann, C., Schönberg, U.: How general is human capital? A task- simulations with global interdependencies. A sensitivity analysis based approach. J Labor Econ 28, 1–49 (2010) for Germany. Econ Syst Res 19(1), 37–55 (2007) Geel, R., Backes-Gellner, U.: Occupational mobility within and be- Montgomery, J.D.: Equilibrium wage dispersion and interindustry tween skill clusters: an empirical anaysis based on the skill- wage differentials. Q J Econ 106(1), 163–179 (1991) weights approach. Empir Res Vocat Educ Train 3, 21–38 (2011) Moscarini, G., Thomsson, K.: Occupational and job mobility in the Gibbons, R., Katz, L.F.: Layoffs and lemons. J Labor Econ 9(4), US. Scand J Econ 109(4), 807–836 (2007) 351–380 (1991) Nisic, N., Trübswetter, P.: Berufswechsler in Deutschland und Großbri- Goos, M., Manning, A., Salomons, A.: Job polarization in Europe. Am tannien. IAB-Kurzbericht 1/2012. (2012) Econ Rev 99(2), 58–63 (2009) OECD: Managing decentralisation – a new role for labour market pol- Groes, F., Kircher, P., Manovskii, I.: The U-shape of occupational mo- icy. OECD Publications Service, Paris (2003) bility. Rev Econ Stud 82(2), 659–692 (2015) Papps, K.L.: Occupational and skill forecasting: a survey of overseas Grossman, G.M., Rossi-Hansberg, E.: Trading tasks: a simple theory approaches with applications for New Zealand. Occasional Paper of offshoring. Am Econ Rev 98(5), 1978–1997 (2008) 2001/1. Labour Market Policy Group, Wellington (2001) Helmrich, R., Zika, G. (eds.): Beruf und Qualiﬁkation in der Zukunft. Pissarides, C.A.: Equilibrium unemployment theory. MIT Press, Cam- BIBB-IAB-Modellrechnungen zu den Entwicklungen in Berufs- bridge (2000) feldern und Qualiﬁkationen bis 2025. Bundesinstitut für Berufs- Pollmann-Schult, M.: Ausmaß und Struktur von arbeitnehmerinduzierter bildung, Bielefeld (2010) Abstiegsmobilität. Kolner Z Soz Sozpsychol 4, 573–591 (2006) Helmrich, R., Maier, T.: Employment forecasting in Germany – an oc- Reichelt, M., Abraham, M.: Occupational and regional mobility as sub- cupational ﬂexibility matrix approach. In: Arendt, L., Ulrichs, M. stitutes. A new approach to understanding job changes and wage (eds.) Best practices in forecasting labour demand in Europe In- inequality. IAB-Discussion Paper 14/2015. (2015) stytut Pracy I Spraw Socialnych, Report II. pp. 103–126. (2012) Schnur, P., Zika, G.: Das IAB/INFORGE-Modell. Ein sektorales Hirschmann, A.O.: The paternity of an index. Am Econ Rev 54(5), 761 makroökonomisches Projektions- und Simulationsmodell zur (1964) Vorausschätzung des längerfristigen Arbeitskräftebedarfs. IAB- Kalinowski, M., Quinke, H.: Projektion des Arbeitskräfteangebots bis Bibliothek 318. Institut für Arbeitsmarkt und Berufsforschung, 2025 nach Qualiﬁkationsstufen und Berufsfeldern. In: Helmrich, Nürnberg (2009) R., Zika, G. (eds.) Beruf und Qualiﬁkation in der Zukunft. BIBB- K 90 T. Maier et al. Caroline Neuber-Pohl is researcher at the German Federal Insti- Shavit, Y., Müller, W.: Vocational secondary education, tracking, and tute for Vocational Education and Training (BIBB), Section “Qualiﬁca- social stratiﬁcation. In: Hallinan, M.T. (ed.) Handbook of sociol- tion, Occupational Integration, Employment”, since 2015. She holds an ogy of education, pp. 437–452. Springer, New York (2000) MSc in Economics, Rheinische Friedrich-Wilhelms-University Bonn, Shaw, J.D., Delery, J.E., Jenkins, G.D., Gupta, N.: An organiza- 2015 and is doctoral student at the Friedrich-Alexander-University, Er- tion-level analysis of voluntary and involuntary turnover. Acad langen-Nürnberg. Her research areas are: Labour market adjustments, Manag J 41(5), 511–525 (1998) Spitz-Oener, A.: Technical change, job tasks, and rising educational de- occupational integration, vocational education. mands. Looking outside the wage structure. J Labor Econ 24(2), Anke Mönnig studied Economics at the Free University of Berlin 235–270 (2006) and holds a Master of Arts Degree in International Economics from Tiainen, P.: Employment forecasting in Finland. In: Arendt, L., Ulrichs, the Berlin School of Economics and Law. In 2006 she joined the In- M. (eds.) Best practices in forecasting labour demand in Europe, stitute for Economic Structures Research (GWS) in Osnabrück. Since pp. 51–62. IPISS, Warsaw (2012) 2016, she is deputy head of the GWS division “Economic and Social Tiemann, M., Schade, H.-J., Helmrich, R., Hall, A., Braun, U., Bott, Affairs”. Her thematic focus concentrates on industrial analysis, labor P.: Berufsfeld-Deﬁnitionen des BIBB auf Basis der KldB1992. markets and structural change with an emphasis on external trade. Wissenschaftliche Diskussionspapiere No. 105. Bundesinstitut She is responsible for the macroeconometric input-output model IN- für Berufsbildung, Bonn (2008) UK Commission for Employment and Skills: Working Futures 2010–2020. FORGE (INterindustry FORcasting Germany) and the world trade Executive Summary, vol. 41. UK Commission for Employment model TINFORGE. and Skills, South Yorkshire (2011) Dr. Gerd Zika studied Business Management at the Friedrich-Alexan- Vogler-Ludwig, K., Düll, N.: Arbeitsmarkt 2030. Eine strategische der University Erlangen/Nuremberg (M. Sc. 1991). Then he worked as Vorausschau auf Demograﬁe, Beschäftigung und Bildung in an assistant at the Chair of Statistics and Econometrics at the Friedrich- Deutschland. Bertelsmann, Bielefeld (2013) Alexander-University Erlangen/Nuremberg (Dr. rer. pol. 1994). In Weber, M.: Wirtschaft und Gesellschaft. Mohr Siebeck, Tübingen 1995 he joined the Institute for Employment Research (IAB), where (1972) Whelpton, P.K.: An empirical method of calculating future population. he is responsible for the BIBB-IAB-Qualiﬁcation and Occupational J Am Stat Assoc 31(195), 457–473 (1936) Field Projections (QuBe-Projekt.de) since 2007. His thematic focus Wilson, R.: Forecasting skill requirements at national and company concentrates on the analysis of both short- and long-term developments level. In: Descy, P., Tessaring, M. (eds.) Training in Europe. Sec- in the labor market, with a main focus on the labor demand side. ond report on vocational training research in Europe 2000. Back- Michael Kalinowski studied Economics at the University of Regens- ground report, pp. 561–609. Ofﬁce for Ofﬁcial Publications of the burg (Dipl. Vw./MSc in Economics). In 2007 he joined as researcher European Communities, Luxembourg (2001) the Fraunhofer Institute for Applied Information Technology (Fraun- Zika, G., Helmrich, R., Kalinowski, M., Wolter, M.I., Hummel, M., hofer FIT) in Sankt Augustin, Germany, research group MikMod. Maier, T., Hänisch, C., Drosdowski, T.: In der Arbeitszeit steckt Since 2016 he is researcher at the Federal Institute for Vocational noch eine Menge Potenzial. Qualiﬁkations- und Berufsfeldpro- Education and Training (BIBB) in Bonn, Germany, in the Section jektionen bis 2030. IAB-Kurzbericht 18/2012. (2012) “Qualiﬁcations, Occupational Integration and Employment”. His re- search interests focus on economics of education and labour market Tobias Maier studied Politics and Management at the University economics including projection of the future qualiﬁcation and occupa- of Constance (Dipl. Verw.-Wiss./Master of Arts). In 2009, he joined tion structure of the population and labour work force. the Federal Institute for Vocational Education and Training (BIBB) in Bonn, Germany, as Researcher in the Section “Qualiﬁcations, Occupa- tional Integration and Employment”. He is responsible for the BIBB- IAB-Qualiﬁcation and Occupational Field Projections (QuBe-Pro- jekt.de) and the econometric estimation and simulation of the yearly supply and demand of vocational training (PROSIMA). His further research areas are: educational development, vocational educational, access to employment and labour mobility. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal for Labour Market Research Springer Journals http://www.deepdyve.com/lp/springer-journals/modelling-reallocation-processes-in-long-term-labour-market-IN8DMYLpJJ

Loading next page...

References (62)

CB Lockard, M Wolf (2012)
Occupational employment projections to 2020. Employment outlook 2010–2020
Mon Labor Rev, 135
PK Whelpton (1936)
An empirical method of calculating future population
J Am Stat Assoc, 31
A Gajdos, E Zmurkow-Poteralska (2014)
Employment forecasts by occupational groups in Poland
Polit Spoleczna Themat, 2014
C Gathmann, U Schönberg (2010)
How general is human capital? A task-based approach
J Labor Econ, 28
G Zika, R Helmrich, M Kalinowski, MI Wolter, M Hummel, T Maier, C Hänisch, T Drosdowski (2012)
In der Arbeitszeit steckt noch eine Menge Potenzial. Qualifikations- und Berufsfeldprojektionen bis 2030
P Schnur, G Zika (2009)
Das IAB/INFORGE-Modell. Ein sektorales makroökonomisches Projektions- und Simulationsmodell zur Vorausschätzung des längerfristigen Arbeitskräftebedarfs
C Dustmann, J Ludsteck, U Schönberg (2009)
Revisiting the German wage structure
Q J Econ, 124
M Kalinowski, H Quinke (2010)
Beruf und Qualifikation in der Zukunft. BIBB-IAB-Modellrechnungen zu den Entwicklungen in Berufsfeldern und Qualifikationen bis 2025
D Acemoglu, D Autor (2011)
Handbook of labor economics
JL Cotton, JM Tuttle (1986)
Employee turnover: a meta-analysis and review with implications for research
Acad Manag Rev, 11
JD Shaw, JE Delery, GD Jenkins, N Gupta (1998)
An organization-level analysis of voluntary and involuntary turnover
Acad Manag J, 41
S Brunow, A Garloff (2011)
Arbeitsmarkt und demografischer Wandel. Anpassungsprozesse machen dauerhaften Fachkräftemangel unwahrscheinlich
GM Cortes (2016)
Where have the middle-wage workers gone? A study of polarization using panel data
J Labor Econ, 34
M Weber (1972)
Wirtschaft und Gesellschaft
M Tiemann, H-J Schade, R Helmrich, A Hall, U Braun, P Bott (2008)
Berufsfeld-Definitionen des BIBB auf Basis der KldB1992
GM Grossman, E Rossi-Hansberg (2008)
Trading tasks: a simple theory of offshoring
Am Econ Rev, 98
C Dustman, A Glitz (2015)
How do industries and firms respond to changes in local labor supply?
J Labor Econ, 33
R Gibbons, LF Katz (1991)
Layoffs and lemons
J Labor Econ, 9
RJ Lucas, E Prescott (1974)
Equilibrium search and unemployment
J Econ Theory, 7
T Maier, R Helmrich (2012)
Creating the initial vocational qualification from the German microcensus
(2003)
Managing decentralisation – a new role for labour market policy
M Reichelt, M Abraham (2015)
Occupational and regional mobility as substitutes. A new approach to understanding job changes and wage inequality
A Ben-Ner, A Urtasun (2013)
Computerization and skill bifurcation: the role of task complexity in creating skill gains and losses
Ind Labor Relat Rev, 66
J Allmendinger (1989)
Educational systems and labor market outcomes
Eur Sociol Rev, 5
B Meyer, C Lutz, P Schnur, G Zika (2007)
National economic policy simulations with global interdependencies. A sensitivity analysis for Germany
Econ Syst Res, 19
T Maier, A Mönnig, G Zika (2015)
Labour demand in Germany by industrial sector, occupational field and qualification until 2025 – model calculations using the IAB/INFORGE model
Econ Syst Res, 27
R Wilson (2001)
Training in Europe. Second report on vocational training research in Europe 2000. Background report
P Böckermann, P Ilmakunnas (2009)
Job disamenities, job satisfaction, quit intentions, and actual separations: putting the pieces together
Ind Relat (Berkeley), 48
E Cottini, T Kato, N Westergaard-Nielsen (2011)
Adverse workplace conditions, high-involvement work practices and labor turnover: evidence from Danish linked employer–employee data
Labour Econ, 18
(2012)
Future skills and demand in Europe. Forecast 2012
M Lapointe, M Charron, G Claveau, M Gendron, EG Gomez, M Grenier, L Ignaczak, J-Y Kim, R Lamy, C Pescarus, N Tremblay-Côté, N Vincent, Y Zou (2008)
Looking-ahead: a 10-year outlook for the Canadian labour market (2008–2017)
H Bonin, M Schneider, H Qunike, T Arens (2007)
Zukunft von Bildung und Arbeit. Perspektiven von Arbeitskräftebedarf und -angebot bis 2020
G Moscarini, K Thomsson (2007)
Occupational and job mobility in the US
Scand J Econ, 109
(2009)
Skills supply and demand in Europe. Medium-term forecast up to 2020
A Damelang, F Schulz, B Vicari (2015)
Institutionelle Eigenschaften von Berufen und ihr Einfluss auf berufliche Mobilität in Deutschland
Schmollers Jahrb, 135
T Maier, G Zika, A Mönnig, MI Wolter, M Kalinowski, C Hänisch, R Helmrich, M Schandock, C Neuber-Pohl, P Bott, M Hummel (2014)
Wages and occupational flexibilities as determinants of the interactive QuBe labour market model
(2011)
Working Futures 2010–2020
N Nisic, P Trübswetter (2012)
Berufswechsler in Deutschland und Großbritannien
AO Hirschmann (1964)
The paternity of an index
Am Econ Rev, 54
U Blien, A Reinberg, M Tessaring (1990)
Die Ermittlung der Übergänge zwischen Bildung und Beschäftigung
Mitt Arbeitsmarkt Berufsforsch, 23
WW Leontief (1951)
The structure of American economy, 1919–1939. An empirical application of equilibrium analysis
CA Pissarides (2000)
Equilibrium unemployment theory
M Abraham, A Damelang, F Schulz (2011)
Wie strukturieren Berufe Arbeitsmarktprozesse? Eine institutionentheoretische Skizze
A Spitz-Oener (2006)
Technical change, job tasks, and rising educational demands. Looking outside the wage structure
J Labor Econ, 24
D Ehing, S Moog (2013)
Erwerbspersonen- und Arbeitsvolumenprojektionen bis ins Jahr 2060
J Labour Mark Res, 46
S Bechmann, V Dahms, N Tschersich, M Frei, U Leber, B Schwengler (2012)
Fachkräfte und unbesetzte Stellen in einer alternden Gesellschaft: Problemlagen und betriebliche Reaktionen
F Groes, P Kircher, I Manovskii (2015)
The U‑shape of occupational mobility
Rev Econ Stud, 82
JD Montgomery (1991)
Equilibrium wage dispersion and interindustry wage differentials
Q J Econ, 106
KU Mayer, GR Carroll (1987)
Jobs and classes: structural constraints on career mobility
Eur Sociol Rev, 3
R Geel, U Backes-Gellner (2011)
Occupational mobility within and between skill clusters: an empirical anaysis based on the skill-weights approach
Empir Res Vocat Educ Train, 3
H Brücker, S Klinger, J Möller, U Walwai (2013)
Handbuch Arbeitsmarkt 2013. Analysen, Daten, Fakten
KJ McLaughlin (1991)
A theory of quits and layoffs with efficient turnover
J Polit Econ, 99
Y Shavit, W Müller (2000)
Handbook of sociology of education
K Vogler-Ludwig, N Düll (2013)
Arbeitsmarkt 2030. Eine strategische Vorausschau auf Demografie, Beschäftigung und Bildung in Deutschland
T Maier, A Afentakis (2013)
Forecasting supply and demand in nursing professions: impacts of occupational flexibility and employment structure in Germany
Hum Resour Health, 11
KL Papps (2001)
Occupational and skill forecasting: a survey of overseas approaches with applications for New Zealand
P Tiainen (2012)
Best practices in forecasting labour demand in Europe
B Fitzenberger, S Lickleder, H Zwiener (2015)
Mobility across firms and occupations among graduates from apprenticeship
M Goos, A Manning, A Salomons (2009)
Job polarization in Europe
Am Econ Rev, 99
M Pollmann-Schult (2006)
Ausmaß und Struktur von arbeitnehmerinduzierter Abstiegsmobilität
Kolner Z Soz Sozpsychol, 4
M Bacharach (1970)
Biproportional matrices and input-output change
B Fitzenberger, A Spitz-Oener (2004)
Bildung

Publisher: Springer Journals
Copyright: Copyright © 2017 by The Author(s)
Subject: Economics; Labor Economics; Sociology, general; Human Resource Management; Political Economy/Economic Policy; Regional/Spatial Science; Population Economics
ISSN: 1614-3485
eISSN: 2510-5027
DOI: 10.1007/s12651-017-0220-x
Publisher site: See Article on Publisher Site

Abstract

J Labour Market Res (2017) 50:67–90 DOI 10.1007/s12651-017-0220-x ARTICLE Modelling reallocation processes in long-term labour market projections 1 1 2 3 1 Tobias Maier · Caroline Neuber-Pohl · Anke Mönnig · Gerd Zika · Michael Kalinowski Accepted: 19 January 2017 / Published online: 5 April 2017 © The Author(s) 2017. This article is available at SpringerLink with Open Access. Abstract Long-term labour market projections are a popu- scenario comparisons. Our results suggest that considering lar tool for assessing future skill needs and the possibility of reallocations but also additionally their dynamics has sub- skill shortages. It is often noted that reallocation processes stantial effects on the projection outcomes. They help draw in the German labour market are hindered due to its strong an insightful picture of the future labour market and prevent standardization and occupational segmentation. However, it over- or understating the potential for labour shortages in is possible that persons leave the occupation for which they several occupations. have been trained for. Disregarding such reallocations and We conclude that the assumptions about how realloca- their dynamics in the projection model is likely to distort tions differ by occupation and to what extent they can be the results and lead to inaccurate practical advice. realized by wage impulses is essential for projection results In this article, we describe for the ﬁrst time, how reallo- and their interpretation. Furthermore, we ﬁnd that in the cations in the labour market can be modelled using occu- German labour market, wage adjustments cannot balance pational ﬂexibility matrices and wage dynamics. Here, it is the labour demand and supply for occupations completely. shown that employers react to labour scarcity by increas- ing wages to attract workers who to some extent can adjust Keywords Labour market · Projections · Germany · their mobility behaviour accordingly. We analyse the aggre- Occupational mobility · Education · Wage development gate impact of this implementation of a reallocation process of labour supply on the projection results by the means of JEL I25 · J20· J21·J22·J23· J24·O11· O15 Modellierung von Anpassungsprozessen in Tobias Maier tobias.maier@bibb.de langfristigen Arbeitsmarktprojektionen Caroline Neuber-Pohl neuber-pohl@bibb.de Zusammenfassung Langfristige Arbeitsmarktprojektio- nen stellen ein beliebtes Analyseinstrument dar, um zu- Anke Mönnig künftige Fachkräftebedarfe und -engpässe aufzuzeigen. moennig@gws-os.com Es wird oft angemerkt, dass gerade der stark standardi- Gerd Zika sierte und beruﬂich segmentierte deutsche Arbeitsmarkt gerd.zika@iab.de Reallokationsprozesse von Arbeitsangebot und -bedarf Michael Kalinowski nach Berufen erschwert. Nichtsdestotrotz sind Wechsel kalinowski@bibb.de aus dem erlernten Beruf keine Seltenheit und müssen bei Federal Institute for Vocational Education and Training, einer langfristigen Projektion nach Berufen berücksichtigt Bonn, Germany werden, sofern keine inadäquaten Handlungsempfehlungen Institute of Economic Structures Research, Osnabrueck, aus vermeintlichen Fachkräfteengpässen und -überschüssen Germany abgeleitet werden sollen. Institute for Employment Research, Nuremberg, Germany K 68 T. Maier et al. In diesem Artikel beschreiben wir erstmals, wie die Im- supply hinges on today’s education attainment. Here, the plementierung eines Reallokationsprozesses durch beruﬂi- occupation represents an institutional link between educa- che Flexibilitätsmatrizen und berufsfeldspeziﬁscher Löh- tion and employment (c.f. Weber 1972; Mayer and Carroll ne stattﬁnden kann. So zeigen wir, dass Arbeitgeber auf 1987; Abraham et al. 2011). In such a market, workers Engpässe durch Lohnerhöhungen reagieren, woraufhin Ar- cannot be regarded as homogeneous and perfectly substi- beitnehmer ihr Mobilitätsverhalten anpassen. Anhand von tutable. The production of different goods or services call Szenarien analysieren wir die Auswirkungen unterschiedli- for different specialized skills and, therefore, not every em- cher Annahmen zur Lohnentwicklung in den Berufen und ployee is suited for every job. This is why, for Germany it is deren Effekte auf das Anpassungsverhalten des Arbeitsan- essential to project occupation-speciﬁc labour demand and gebots. Unsere Ergebnisse zeigen, dass sich die Berück- supply in order to yield insightful results (Lapointe et al. sichtigung beruﬂichen Mobilitätsverhaltens sowie eine dy- 2008; CEDEFOP 2012; Helmrich and Zika 2010). namische Entwicklung desselben substanziell in den lang- However, although these submarkets are linked to a spe- fristigen Projektionsergebnissen niederschlagen. Hierdurch ciﬁc occupation, they are not totally restrictive. The trans- ergibt sich ein differenzierteres Bild über mögliche Fach- ferability of task-based human capital enables occupational kräfteengpässe und -überhänge sowie mögliche Handlungs- mobility to related ﬁelds (Gathmann and Schönberg 2010). empfehlungen. In fact, Nisic and Trübswetter (2012) calculate that ev- Als Fazit lässt sich festhalten, dass mögliche Lohnanpas- ery year about 3.4% of Germany’s employed population sungen und damit verbundene Berufswechsel zu einem bes- change their occupation. To put this into perspective, Nisic seren Ausgleich von Arbeitsangebot und -nachfrage nach and Trübswetter (2012) calculate a yearly rate of 10.8% Berufen führen können und dass Annahmen über den Ab- in Great Britain. For Denmark, Groes et al. (2015)talk of lauf dieser Prozesse das Ergebnis stark beeinﬂussen. Zudem a yearly occupational mobility rate of 20% and Moscarini können wir für den deutschen Arbeitsmarkt konkludieren, andThomsson(2007) estimate a monthly rate of 3.5% dass nicht alle theoretischen Engpässe lediglich über Lohn- among male workers in the US. Thus, in the international erhöhungen lösbar sind. comparison, a yearly rate of 3.4% may actually be a rel- atively small number. Nevertheless, this level of mobility can to a certain extent be thought to resolve misallocations 1 Introduction of the working population. Furthermore, disregarding the opportunities and limitations of occupational ﬂexibility and The German economy and labour market are subject to its dynamics in projection models is likely to distort the structural change over time. Demographic change, techno- results (cf. Brücker et al. 2013; Brunow and Garloff 2011). logical progress, and globalisation will frame the behaviour Notwithstanding, projection models have to trade-off of market participants. Political planners have a special in- transparency of results and accuracy to some extent; ac- terest in having some knowledge about the future – be it curately reﬂecting all underlying mechanisms may cause for budgetary planning or preliminary policy assessments. separate effects not to be identiﬁable and results not inter- In addition, regarding future developments of the labour pretable (Wilson 2001). Therefore, the decision of whether market, a concern is whether the supply of skills will suf- or not and how to implement reallocation dynamics in ﬁce the demand of the economy, such that growth can a projection model of the German labour market is not spur, or whether there is a possibility of labour shortages. trivial. Here, long-term labour market projections are a more and Helmrich and Zika (2010) for the ﬁrst time model oc- more popular tool for policy consulting (Wilson 2001). To- cupational ﬂexibilities into a long-term projection of the day many countries have such projections (cf. for example German labour market, the BIBB-IAB qualiﬁcation and CEDEFOP 2009 and 2012 for Europe; Dupuy 2012 for occupational ﬁeld projections (QuBe, henceforth). Based the Netherlands; Gajdos and Zmurkow-Poteralska 2014 for on this, Maier et al. (2014) propose a dynamic realloca- Poland; Bonin et al. 2007; Maier et al. 2014; and Vogler- tion mechanism for the qube model, which redistributes Ludwig and Düll 2013 for Germany; Lapointe et al. 2008 labour supply to labour demand via occupational mobil- for Canada; Lepic and Koucky 2012 for the Czech Repub- ity given wages. This is a novel approach to model long- lic; Lockard and Wolf 2012 for the US; Tiainen 2012 for term projections and to our knowledge has not been done Finland; Papps 2001for New Zealand; UK Commission for in any other labour market projection so far. In the model Employment and Skills 2011 for the UK). of Maier et al. (2014), employers respond to occupation- Especially in Germany, where the labour market is highly speciﬁc labour scarcity by raising wages, which in turn segmented into occupation-speciﬁc submarkets (cf. Mayer causes trained workers and workers from related disciplines and Carroll 1987; Allmendinger 1989; Shavit and Müller to more often offer their work in this occupational submar- 2000;OECD 2003), the balance of the labour demand and ket. In this paper, we wish to highlight the impact of this K Modelling reallocation processes in long-term labour market projections 69 modelling approach on the QuBe projection results and the effect on the projection results. In this section, to start with, overall importance of considering reallocation mechanism we will brieﬂy discuss the choice of a purely wage driven in labour market projections in the context of the evaluation mechanism reﬂecting on related literature on the topic of of possible hazards of labour supply shortages in the future. turnover, employer recruitment strategy, and the drivers of Our analysis will show in which occupations, we can rely occupational mobility in general. on market mechanisms to solve possible labour shortages via wage dynamics and in which occupations, enterprises 2.1 The employer’s adjustment mechanism and policy makers have to intervene by for example im- proving working conditions in general or providing further Projection results are often said to exaggerate the extent educational training. of possible labour shortages in the future. This critique of- In the following, we ﬁrst discuss whether wage-based ten addresses that adjustment mechanisms of employers are dynamics of the reallocation process are adequate by re- neglected in the analysis (cf. for example Brücker et al. viewing recent literature on this topic (Sect. 2). In the 2013). Brunow and Garloff (2011) even reject the idea of third section, we brieﬂy give an intuitive introduction to future labour market shortages in total. They argue that in the QuBe model and describe its reallocation mechanism in the event of a tightening labour market, employers have more detail. In the fourth section, we outline the different plenty of ways to adjust adequately and prevent a shortage data sources used for the QuBe model and how the real- situation. They suggest that ﬁrms will react to the antici- location dynamics where operationalized. Sect. 5 presents pation of a shortage by substituting their labour demands results from scenario comparisons, which illustrate the ef- by automating processes or hiring workers from abroad. fect of this modelling on the projection results. Here, we Also ﬁrms could alter their stock of capital and produce ﬁrst assess the overall impact of implementing the reallo- less, thereby demanding less labour. Brunow and Garloff cation mechanism in QuBe (Sect. 5.1). Then we show how (2011) also highlight the importance of wages, which they the dynamic adjustment of employers and workers to each consider ’upward ﬂexible’ enough to attract the necessary other take a great part in the overall effect (Sect. 5.2). After labour supply. this, we discuss how the interpretation of the results are Economic theory, likewise, predicts a relationship be- strongly inﬂuenced also by the implicitly modelled limita- tween wages and relative labour supply. Especially in the tions of wage dynamics in balancing the labour market by search and matching literature labour market tightness ex- presenting results from wage policy scenarios (Sect. 5.3) plicitly enters the wage equation such that a shortage of and discussing to what extent the calculated optimal ﬂexi- applicants always corresponds to higher wages (cf. for ex- bility of the workforce is achievable via the wage mecha- ample Pissarides 2000). Montgomery (1991), for example, nism (Sect. 5.4). In Sect. 6, we conclude and give an outlook uses a related model set-up to explain wage differences on future research. across industries. Here, ﬁrms who value ﬁlling their va- cancy most, pay the highest wage in order to overcome coordination problems and attract the most applicants to 2 Theoretical assumptions and related empirical their opening. ﬁndings However, Bechmann et al. (2012) show that wage policy may be less important to German recruiters. They analyse In order to account for reallocation dynamics in their pro- data of the IAB Establishment panel , where ﬁrms were jection model Maier et al. (2014) let employer-set wages asked which strategies they used or would use to allevi- partially depend on labour supply scarcity. Labour supply, ate labour shortages. The most important strategy, in fact, in turn, responds to differences in relative wages of occu- seems to be further training of the current workforce, which pations by changing their occupational mobility behaviour was chosen as very important by 42% of the surveyed ﬁrms. in that the workers propensity to stay in their training oc- Next to other means of recruiting from within the company, cupation correlates positively with a lower outside option. as for example later retirements or apprenticeship programs, In this model set-up, wage is the only explicit adjustment also the attractiveness of the job offer was stated to be tar- channel of employers and worker behaviour in response to geted. With 34% of the establishments highlighting its im- misallocations of labour. All other factors, which inﬂuence portance making the offer desirable seems to be the second mobility decisions of workers, are assumed to follow a con- most important strategy of ﬁrms. In contrast, wages seem stant time trend. Other factors, which drive wage setting of to be less important. Only 11% of the ﬁrms consider pay- the employer, are assumed to relate to the production pro- cess and outside wage pressures. The Establishment Panel of the Institute for Employment Research In the following sections, we will describe this mecha- (IAB) representatively surveys about 16,000 German establishments nism in more detail and outline its empirical foundation and on their employment policies and related topics since 1993. K 70 T. Maier et al. ing higher wages as an important strategy. It is, however, voluntary lay-offs are associated with a switch to lower still a strategy for 47% of the surveyed ﬁrms, even though wages (McLaughlin 1991), which Gibbons and Katz (1991) 36% indicate that a main problem concerning recruiting is, explain with the ‘lemon effect’ causing laid-off workers in fact, too high wage demands of applicants (Bechmann having troubles with ﬁnding a new job. The importance et al. 2012). of the nature of the switch is also highlighted by recent Eventually, Dustman and Glitz (2015)and Dustman et al. results of Fitzenberger et al. (2015). Providing evidence (2009) ﬁnd empirical evidence for the impact of the struc- concerning the occupational mobility of recent apprentice- ture of skill supply on wages. Using IAB Establishment ship completers in the German labour market, they ﬁnd that Panel data from 1985 and 1995, Dustman and Glitz (2015) mere job switches inside the occupation but between ﬁrms investigate whether employers in West Germany react to most often lead to a wage loss, while occupational mobil- a change in the skill mix of the workforce by adjusting ity is associated with a wage gain in most cases. However, wages or the production intensity, where they distinguish they point out that occupation-and-ﬁrm switches only result between switching to production of goods, which can be in a gain if this switch reﬂects an occupational upgrading, produced by the skills available, or producing the same while occupation switches within the ﬁrm, which reﬂect goods but adjusting the skill application. They conclude a switch to a better ﬁtting position, are usually associated that ﬁrms adjust mainly by the latter. Concerning wage ad- with awagegain. justments, they ﬁnd that wages are only signiﬁcantly elastic Other research points toward the increasing wage in- with respect to skill supply in the nontradable and manu- equality. Groes et al. (2015) point out that mainly low and facturing sector, where a 1% increase of skill supply corre- high income earners switch occupations and that downward sponds to a 0.4% and 0.1% decrease in wages, respectively. mobility seems to be a phenomenon of low income earners. Dustman et al. (2009) come to a similar conclusion. Taking An explanation for this, according to Groes et al. (2015), is advantage of the change in skill structure of the German that occupations with rising productivities layoff their low labour market induced by the reuniﬁcation, they show that skilled workers (and typically low wage earners), leaving the relative abundance of lower skilled workers after the in- them to seek work in other occupations, while high skilled tegration of the East German Länder increased skill returns. workers move out of the declining productivity occupations To sum up, there is evidence of ﬁrms reacting to labour in order to obtain higher wages. As a result, again only the market tightness by raising wages in order to attract suf- high skilled workers are hypothesized to experience wage ﬁcient applicants to their vacancies. However, the extent increases when switching their occupation. of the wage mechanism may be relatively small as ﬁrms The literature on task biased technological change ex- also use other strategies to overcome recruitment problems. plains the observed trends in wage inequality by job polar- These include training and solutions for better working con- ization. Emerging new technologies, which automate many ditions (cf. Bechmann et al. 2012). routine tasks, and globalisation, which poses new opportu- nities for offshoring (see also Grossman and Rossi-Hans- 2.2 The worker’s adjustment mechanism berg 2008), cause redundancy of domestic labour in some occupations (see for a summary Acemoglu and Autor 2011; In labour economics, there has been a long debate about Goos et al. 2009). Such a trend can also be found for Ger- whether job or occupational mobility is associated with many (cf. Spitz-Oener 2006). Cortes (2016) explains this a wage gain or a penalty. The classic island model by Lu- polarization effect further by the induced sorting on ability cas and Prescott (1974) would predict that negative demand among the workforce. According to this, more able workers shocks motivate workers (low skilled ﬁrst) to leave their job will sort into occupations with higher non-routine, cognitive to seek higher wage opportunities. Likewise the search and task shares, while less able workers switch to high routine, matching literature (c.f. Pissarides 2000 for an overview) non-cognitive jobs. Therefore, only the more able workers predicts a positive relationship between job mobility and will experience a rise in wages upon a job switch. outside wages, as workers are rational and only move if in- Yet another interpretation for the duality of wage out- centivized. For the German labour market, Fitzenberger and comes upon occupational changes is presented by Gath- Spitz-Oener (2004) ﬁnd an overall positive relationship be- mann and Schönberg (2010) and also Geel and Backes- tween occupational switches and wages, thereby supporting Gellner (2011). They attribute the probability of a wage that occupational mobility mainly serves as a career seeking gain after a switch to the proportion of speciﬁcity of the device. acquired skills in the former occupation. Geel and Backes- However, there is also always a non-negligible share of Gellner (2011) show that the higher the speciﬁcity of skills, job switchers who have experienced downward mobility (cf. the lower occupational mobility. In addition, Gathmann Gibbons and Katz 1991). Whereas voluntary quits are most and Schönberg (2010) also show that occupational mobil- often associated with separations to higher paying jobs, in- ity mostly entails switches to related ﬁelds, where skills are K Modelling reallocation processes in long-term labour market projections 71 best transferable. Apart from the share of speciﬁc human tute for Employment Research (IAB) in collaboration with capital needed in an occupation, Damelang et al. (2015) the Fraunhofer Institute for Applied Information Technol- indicate that also to the degree of standardisation and occu- ogy (FIT) and the Institute of Economic Structural Re- pational closure is important. A higher degree of regulation search (GWS). As this paper focuses on possible reallo- (meaning the existence of occupation speciﬁc VET certiﬁ- cation mechanisms of labour demand and supply to over- cates and study programs) reduces the propensity of leaving come long-term mismatches at the occupational level, we the occupation. will only brieﬂy touch on the derivation of labour demand Additionally, there are of course also other factors driv- and supply in the QuBe projections and describe the im- ing job mobility aside from monetary incentives. Cotton and plemented reallocation mechanism more thoroughly. The Tuttle (1986), Shaw et al. (1998), Pollmann-Schult (2006), reader is referred to Maier et al. (2014, 2015) for a detailed Böckermann and Ilmakunnas (2009), Cottini et al. (2011) description of the model. Note that the working volume is all emphasize the importance of physical and psychological central to the demand side model and results are also avail- hygiene, as well as, a good work life balance for retention of able in aggregate hours of work. However, for simplicity in employees. Furthermore, on more regional level, regional this paper we only focus on results evaluated in the number mobility within an occupation has to be considered as an of persons involved. alternative to occupational mobility (Reichelt and Abraham The underlying model projects a development path (the 2015). baseline scenario) of the German economy into the future Furthermore, note that other mechanisms that do not con- given that the currently observable behavioural patterns and cern occupational mobility may also be used in projection trends in the goods, labour and education market will con- models. Ehing and Moog (2013) point out that the size of tinue on their develop path until 2030. As such, it does the future workforce hinges on assumptions about future not necessarily represent the most likely development, but labour force participation. Zika et al. (2012) suggest that can be understood as an outlook on the possible structure the amount of hours a person wishes to work signiﬁcantly of the future labour market when every market participant impact labour supply, especially in occupations with large keeps on her current path of motion. Using this approach shares of part-time workers. This suggests that one could enables a straight forward interpretation of the results and also implement a mechanism, which assumes workers to makes them easily comparable to outcomes of alternative react to changes in the labour market by altering their par- scenarios. In this spirit, modes of behaviour, which cannot ticipation or their working volume. Also migration ﬂows be empirically veriﬁed, are considered infeasible for the re- could dynamically adjust to the labour market situation in sulting baseline scenario. Thus, for example technological a projection model. However, such mechanisms have not progress is only captured by a constant trend and not as- been implemented in any projection model so far. In the sumed to accelerate until 2030. We do, however, implement QuBe model all of these measures are assumed to be stable future changes which have been enacted by legislation and or to follow a trend in their development. have a relevant effect on the outcome during the projection To sum up, in theory wage impulses should create an period. As an example, the baseline scenario takes the new incentive to switch occupations. However, not all occupa- German pension age of 67 into account. tional switches are found to be associated with an increase Fig. 1 gives a highly simpliﬁed overlook of the QuBe in wages. Therefore, in the aggregate the effect of wages on model. Two concurrent processes essentially determine occupational mobility may be mediated by downward mo- labour market outcomes: The evolution of labour supply bility of a part of the occupation switchers. Indications that driven by demographic change (left box) and the evolution the possibility of downward movements is associated with of labour demand, which is driven by economic structural the nature of the task or the prior income level, suggest that change (right box). Both labour supply and demand devel- wage effects should differ by occupations. In addition, other opments are projected until 2030. Essential to the model factors concerning the perceived attractiveness of the occu- is the distinction between the training occupation, which pation seem to have an important impact of occupational workers are associated with on the supply side, and exer- mobility. cised occupation, which workers relate to on the demand side of the labour market. On the supply side, we project the numbers of new labour 3 The BIBB-IAB qualiﬁcation and occupational supply, those leaving the labour market, and ultimately the ﬁeld projections total supply given their sex, age, qualiﬁcation level, and training occupation. For this purpose, the Fraunhofer FIT In this section, we will describe the underlying model. The developed a cohort component model (c.f. Whelpton 1936; QuBe model is a joint project of the Federal Institute for Blien et al. 1990; more speciﬁcally for QuBe see also Kali- Vocational Education and Training (BIBB) and the Insti- nowski and Quinke 2010), which subdivides the popula- K 72 T. Maier et al. Fig. 1 The QuBe model (Source: QuBe projections; rd 3 wave) tion according to sex, age, and qualiﬁcation characteristics not taken into account. While the short-term may be con- and extrapolates the in- and outﬂows of these subgroups cerned with, for example, dealing with the consequences into the future (BIBB-FIT model). The movements between of the euro crisis, structural change is the essential deter- groups summarize ageing given births and deaths, migra- minant of labour demand in the long-term. In pursuance of tion, and qualiﬁcation attainment behaviour. The latter is accurately reﬂecting structural change, QuBe relies on the simulated with a nested transition model of the German QINFORGE model developed by the GWS – a further de- education system. Here, pupils are allocated and transition- velopment of the IAB-INFORGE model (Meyer et al. 2007; ing between high school tracks, entering the vocational ed- Schnur and Zika 2009; Maier et al. 2015). QINFORGE is ucation system, switching between higher education and an econometric input-output model for Germany, which is vocational training programs and, ﬁnally, according to the overall completion rates of the different programs ﬁnishing 2 Vacancies are not taken into consideration in the QuBe long-term by obtaining a credential assigning them to a qualiﬁcation projections for four reasons:. Micro-macro problem: At the micro-economic level, the non-ﬁlling level and according to the prevailing empiric rates of oc- of a vacancy leads to a loss if it causes the company concerned to refuse cupation attainment a training occupation, which they can orders and, thus, to restrict or not to expand production capacity. This use in the labour market to earn wage proﬁts. Of course, does not, however, necessarily mean that there is a corresponding loss infeasible transitions which cannot be identiﬁed in the data in production for the economy as a whole, i. e. at the macro-economic level. Indeed, it may instead lead to the acceptance of the order by are not considered. Note further that people in or without another domestic company, which instead expands its production ca- any vocational education do not have a training occupation pacity, offsetting the potential loss in demand. by deﬁnition and can, therefore, only be associated with an Methodology: Without further background knowledge, no expansion exercised occupation if they are economically active. The demand can be deduced solely from an increase in vacancies, since the number of vacancies cannot be differentiated according to replacement number of economically active persons for each subgroup and expansion demand. is calculated using group speciﬁc participation rates, which Long-term observation: From an economic point of view, vacancies are forecasted with a logistic trend model. only become a problem – if at all – if they cannot be ﬁlled. Even if On the demand side, we calculate the total number of we do not impute complete information or rational agents, problems with an unﬁlled vacancy should vanish with time as a result of the persons needed to manufacture and provide the total num- reallocation process. Therefore, we safely that the number of vacancies ber of goods and services produced in Germany given their always returns to its frictional level in the long term. qualiﬁcation and exercised occupation for each economic Data quality: Reported vacancies statistics by the Federal Employ- sector. We refer to this as realised demand; vacancies are ment Agency (BA) also contain vacancies that do not have to be ﬁlled necessarily. The reasons for this may be multifarious: neglect of report- ing a successful ﬁlling by the company or duplicate reports. Although this problem does not arise with data of the Job Vacancy Survey con- ducted by the IAB, the data here is not available to a sufﬁcient depth of occupational disaggregation. K Modelling reallocation processes in long-term labour market projections 73 ld deeply disaggregated by economic sectors and commod- w = ˛ + ˛ W + ˛ (1) o 1 2 3 ls ity groups. To describe this model in a very simpliﬁed way, let state, employers, and private households invest In a further step, the industry- and occupation-speciﬁc and consume, thereby generating demand. On top, there wage (w ) is modelled. Here, note that the QuBe model as- o;i is a demand for German products from abroad. Also, in- sumes an underlying productivity-based wage policy. Thus, ternational trade poses price pressures on exports and im- industry level wage differences within occupations are ex- ports, which affect price levels for consumption but also plained by differences in labour productivity. Thus, production goods in Germany. This affects the demand for imported goods and also raises unit costs for German prod- w = ˇ + ˇ w + ˇ lpp ; (2) o;i 1 2 o 3 i ucts. Given the individual input-output interdependencies of the economic sectors, the production level is raised or low- where lpp denotes the industry speciﬁc productivity of ered accordingly. Production results in value creation and labour. Again, a constant is included to account for any employment, leading again to a reaction of consumption time invariant determinants of the level of industry- and and investments. In an iterative process these described in- occupation-speciﬁc wages. terdependencies between the different economic actors de- After modelling the wage dependency on labour scarcity, termine the ﬁnal growth path of Germany and the level of the industry and occupation speciﬁc wage is integrated into employment per economic sector, which, according to the the projection of labour demand. Demand for labour by structure of each sector, translates to a demand of labour occupation and industry is explained by the relative appli- for each exercised occupation. cation of the occupation in the economic sector as given Having derived both labour demand and supply, we con- by its contribution to total industry volume of work, i. e. tinue now with a more detailed description of the reallo- occupation- and industry-speciﬁc volume of work relative cation mechanism, which connects both sides (see Fig. 1). to total industry volume of work. The industry-speciﬁc vol- Sect. 3.1 will be concerned with the wage adjustment mech- ume of work is driven by the output level and constraint anism of employers, while Sect. 3.2 will outline the oc- by industry-speciﬁc wage costs. Also, due to technological cupational ﬂexibility adjustment mechanism of workers. progress it is explained by a decreasing time trend indicat- Together both mechanisms form the reallocation process ing the growing efﬁciency of labour inputs. The connection imbedded in the QuBe model. However, we wish to point between volume of work and labour scarcity is modelled out that such a reallocation mechanism could easily be by Eq. 3. transferred to other projection models. vow w o;i o;i = + + t (3) 1 2 3 vow w i i 3.1 Modelling wage adjustment due to skill shortages The equation states that the relative differences in work This section describes the labour demand adjustment mech- inputs between occupations in the same industry is ex- anism through the wage channel with respect to labour mar- plained by a time trend (t) and the relative wage difference o;i ket tightness. Note that the occupation dimension to a very ( ). The latter depends on the occupation speciﬁc labour high extent already captures the informational input of qual- scarcity (cf. Eq. 1). Thus, relatively scarce labour will be iﬁcation. relatively pricy such that its application in the production The starting point is the occupation speciﬁc wage, which process measured by its volume of work is lowered. Given is a function of the total average wage in the economy (W ), that the amount of annual hours worked by one labourer and a scarcity term. The latter is given by the ratio of labour in this industry and occupation does not change, there will demand (ld ) and supply (ls ) in the occupation and op- be a decrease in labour demand in this occupation in this o o erationalizes the overall tightness within the occupational industry. Note that an adverse shock to scarcity causes a per- submarket. W itself is a function of aggregate per capita turbation, since the resulting change in labour demand will labour productivity, overall ﬂuctuation in prices and an ag- in turn alter the scarcity measure again, which moderates gregate term of the labour market tightness for the entire wages and labour demand. Such a perturbation also affects economy. Additionally, a constant is included, which cap- other industry wages through a change in aggregate income. tures all occupation-speciﬁc time invariant factors, which This modiﬁes consumer demand, which is the main driver also determine occupation wages. This captures, for exam- for production in a lot of industries. An increased produc- ple, the extent to which employers could overcome labour tion level induces a raise in labour demand, which again shortages by raising employee productivity by innovative starts off the process of wage adjustments in the affected technologies or further training within a certain occupation industries. (cf. Sect. 2.1). K 74 T. Maier et al. 3.2 Modelling occupational ﬂexibility due to wage 4.1 Data and classiﬁcations adjustments For the QuBe model, data from a number of sources This section outlines the reallocation process of labour sup- was merged to generate a unique data set, which outlines ply on the occupational level through the wage channel. The a deeply disaggregated picture of the German economy and basic idea is that within the model occupational switches are the labour market. For structural information, we rely on accounted for, i. e. it is not assumed that a person who has data of the years 1996 to 2011 retrieved from the German been trained in a certain occupation automatically is part of Microcensus (Labour Force Survey), which is a yearly sam- this occupation-speciﬁc labour supply. Therefore, the start- ple survey of roughly 1% of the German households. It is ing point of modelling this mechanism is the distribution the main source of information for the population structure of the skilled labour force by training occupation over all with regard to age, sex, qualiﬁcation level, employment sta- exercised occupations. Persons, for which the training and tus and training occupation (Maier and Helmrich 2012). It the exercised occupation are identical, are called stayers, also provides data on the distribution of gainfully employed henceforth. The share of stayers in the training occupation, persons over industries and exercised occupations for the to, is denoted by st aye r . years 2005 to 2011 and can, therefore, also be used to anal- to This stayer share is assumed to be time variant and reacts yse occupational switches. Furthermore, it contains data on to impulses of the economic environment. In the model, self-employed and civil servants. No other survey delivers these impulses are captured by outside wage opportuni- a more complete picture for all these characteristics. ties given by a training occupation speciﬁc reference wage On the demand side, information on consumption, prices, ref (w ), which is the weighted average of the wages of all and production for the years 1991 to 2011 is retrieved to (inside and outside) work opportunities, which are feasible from the National Accounts of the Federal Statistical Of- (considering the distribution over exercised occupations) ﬁce (FSO, henceforth). Especially, the input-output-tables for a certain training occupation. The share of stayers is enable a modelling of the interindustry dependencies within determined by equation the production process. For the wage development, we retrieve daily wages for to st aye r = ı + ı (4) full-time employees of the years 1993 to 2011 from the to 1 2 ref to IAB Employment History Data (EHD), which records all where w denotes the wage in the training occupation, to. employment relationships subject to social security contri- to The equation states that whenever a certain training occupa- butions in Germany and captures information about work- tion experiences an increase in wages while the wage level ing days per person and wage totals by economic indus- remains constant in all other reference occupations, it will try, occupation exercised and qualiﬁcation level. By relying become relatively more proﬁtable to stay in the training oc- on this data set, note that we misrepresent wages of civil cupation, thus, causing a rise in the share of stayers. The servants, self-employed and helping family members. Also, extent to which the intent to stay in the training occupa- wages of top income earners are underestimated due to legal tion reacts to outside wage pressures is determined by ı , censorship in the upper income range. However, employees which is the training occupation-speciﬁc wage elasticity of subject to social insurance contributions represent the ma- the propensity to stay. Again, a wage rise triggers a pertur- jority of the work force (about 89% in 2015) and there is bation, where the aggregate effects on labour supply cause no larger and more detailed dataset on gross wages avail- a re-evaluation of wages and labour demand, which, in turn, able in Germany. We, therefore, use the wage development causes preceding adjustments of the supply side and so on. of the EHD as indicator for the general occupation and in- dustry speciﬁc wage development. Note also, that with the underlying data the new legislation on minimum wages is 4 Operationalization and estimation of the QuBe not yet accounted for. model Furthermore, we use the 12th Coordinated Population Forecast of the Federal Statistical Ofﬁce ‘Version 1–W2: In the following section, we brieﬂy present the data used Upper limit of the “medium” population’ until 2060 to to estimate the QuBe model and point out some indication quantify the population by age and sex in the future. To of the explanatory power of scarcity for labour demand and wages for labour supply, respectively, before we fur- A preliminary assessment of the minimum wage policy based on the th QuBe model was presented on the 11 International Conference Chal- ther highlight the magnitude of their impact by sensitivity lenges of Europe in 2015. The results suggest a negative overall impact analyses in the subsequent section. on the economy. Service-oriented industries and professions with low to medium-skilled qualiﬁcations are likely to be exposed the most. See also URL: https://www.efst.hr/eitconf/index.php?p=proceedings. K Modelling reallocation processes in long-term labour market projections 75 rd be able to account for the current developments in the pop- est available Microcensus when the 3 wave of the QuBe ulation in both absolute terms and in terms of their changed project was computed. Secondly, it was also the last Mi- age structure, Version 1-W2 was adapted to the new results crocensus, which used the KldB92 to classify occupations. of the Census 2011. Note that Version 1-W2 is meant to re- Thereafter a harmonization of past data to the 2010 Classi- ﬂect an upper limit of the population, however, understates ﬁcation of Occupations is needed. the current net migration inﬂows of, in particular, political and religious refugees. Accounting for this is likely to im- 4.2 Estimation of the QuBe model pact the projection outcomes. As an example, the demand for teachers may be increased considering the high share In this section, we brieﬂy outline how the before mentioned of young migrants. Therefore, the QuBe projection results, equations of the reallocation mechanisms were estimated. as well, are outdated in this sense. This illustrates how the Using data from 1993 to 2011 on daily wages of full- plausibility of long-term projections strongly hinges on cur- time employees, working volume and labour productivity, rent beliefs of future developments. However, to show the Eqs. 1 to 3 were estimated adding an error term to the right effects of different modelling assumptions concerning the hand side, where the subscripts o and i are captured by the adjustment process on the projection results it can also be 54 OF and the 63 aggregated economic sectors, respectively. helpful to isolate effects from such factors. We, therefore, The t-test for the parameters of Eq. 1 indicate (at a signif- think that our results can be used to visualize the impact of icance level of 5%) that the measure of labor scarcity is the modelling of the reallocation process, even though the a good, necessary and observable predictor for wage level recent migration behaviour is not taken into account. differences between occupations. Especially for ‘occupa- For the calculation of new labour supply by qualiﬁca- tions concerning the production of chemicals and plastics tion level and formal vocational qualiﬁcation, the forecasts wages’ largely, signiﬁcantly depend on labour market tight- of the Conference of Ministers of Education and Cultural ness. However, in 8 of the 54 OF, the effect of scarcity is Affairs of the Länder in the Federal Republic of Germany found to be insigniﬁcant. An example is the ‘public admin- of pupils and graduates from German high schools and uni- istration occupations’. An explanation could be the lack of versity entrants until 2025 are used as a benchmark for the variation in the scarcity variable in these OF. future development in schools and higher education. The Eq. 2 uses the results of Eq. 1 for estimating occupation- retrieved entry, graduation and transition rates for 2025 are speciﬁc wages in each of 63 industrial sectors. A potential held constant thereafter until 2030. of 3402 wages are estimated accordingly. However, not all For both the supply and the demand side the date is ag- occupation and industry combinations exist: taking 2010 gregated using the same classiﬁcation schemes. The Inter- as base year, only 75% of all possible combinations report national Standard Classiﬁcation of Education 1997 is used employment. The corresponding regressions are estimated to differentiate between qualiﬁcation or skill levels. For using ordinary least squares. The estimated parameters are the occupation dimension, the 369 occupational categories evaluated against the R (greater than 0.90), Durbin-Wat- (3-digit code) of the 1992 Classiﬁcation of Occupations son test statistic (between –1 and 1), and the p-value (be- (KldB92) are aggregated according to the 54 occupational low or at 0.05). In total, it was possible to identify wage ﬁelds (OF, henceforth) of Tiemann et al. (2008). Using the responsiveness in 1.513 occupation-speciﬁc industry wages OF to distinguish between occupations prevents artefacts which means that roughly 30 thousand employees are wage- in the modelling of occupation switches, which particularly sensitive in an econometric sense. Nonetheless, there exist occur in the manufacturing sectors because the KldB92 is some cases for which no conclusions about the existence of very detailed here. For an easier visualisation, we report an industry-speciﬁc penalty or mark-up can be made, be- our results for 20 main occupational ﬁelds (MOF, hence- cause either the coefﬁcient of the industry-speciﬁc labour forth) – an aggregated version of the OF (see Table 5 in productivity is insigniﬁcant or the regression is subject to the appendix). Economic sectors are classiﬁed using the autocorrelation. In these 28% of the cases a default option aggregation to 63 industries of the National Classiﬁcation is used, using the OF wage to update the industry speciﬁc of Economic Activities of 2008 (Table 6 in the appendix). OF wages. A similar approach is used for the estimation of To harmonise the supply and demand side data, the num- Eq. 3, where in cases of autocorrelation or insigniﬁcance of ber of persons in active employment as retrieved from the the wage relation by default the relative inputs of occupa- Microcensus is re-extrapolated to match the total number tions is kept constant. Therefore, not in all cases changes in as recorded in the National Accounts, while retaining the the labour supply transmit a change in wages and likewise structure of the population by age, sex, educational level not all wage changes induce a change in the occupational and formal vocational qualiﬁcation from the Microcensus. structure of the industry. For the estimation of Eq. 4, ﬁrstly, Throughout, 2011 is the base year of the QuBe projection. the distribution of formally trained workers by 54 training The reason is that ﬁrstly, the Microcensus 2011 was the lat- OF over the exercised OF is calculated for each age, sex K 76 T. Maier et al. Table 1 Occupational ﬂexibility matrix from formal vocational qualiﬁcation to occupation exercised in 2011 for 20 MOF MOF formal vocational Switches to MOF exercised (in %) qualiﬁcation 1234 5 6 789 10 11 12 13 14 15 16 17 18 19 20 Total 1 Raw material pro- 51.1 3.1 1.6 3.5 0.6 2.9 3.2 2.5 10.9 1.2 2.1 4.2 4.9 0.9 1.2 2.6 0.9 1.2 0.8 0.6 100 cessing occupation 2 Auxiliary workers, 0.0 66.3 6.6 6.9 0.0 0.0 0.0 1.4 2.6 1.7 0.0 1.7 2.6 0.0 2.9 4.6 1.3 0.0 0.0 1.5 100 janitors 3 Metal production 1.4 5.8 36.6 3.6 1.3 9.8 1.3 2.6 10.2 2.2 1.1 2.0 3.8 3.4 8.5 3.5 1.4 0.7 0.4 0.5 100 and processing, in- stallation, electrical occupations 4 Construction, wood- 2.1 5.8 3.2 46.9 1.9 5.5 1.7 2.0 14.0 2.3 1.1 2.8 2.7 1.0 2.1 2.3 0.9 0.8 0.5 0.5 100 working, plastic manuf. occupations 5 Other processing, 1.8 4.1 4.3 3.4 25.3 7.6 5.4 3.7 16.6 2.1 3.3 7.2 4.6 0.9 2.9 2.6 1.2 1.6 0.7 0.6 100 producing and main- taining occupations 6 Machinery and 1.2 3.5 7.4 2.7 3.1 41.3 2.0 2.5 10.8 2.1 1.6 3.5 3.9 1.8 5.2 2.7 2.8 0.9 0.5 0.5 100 equipment steering/ maintenance occup 7 Commodity trade in 1.4 1.9 1.1 0.2 0.2 1.1 49.5 4.3 5.9 0.5 6.0 12.1 7.9 0.2 0.2 1.6 0.8 3.5 1.3 0.3 100 retail 8 Commodity trade 0.6 1.2 0.6 0.3 0.3 0.9 13.6 34.6 5.4 1.1 3.3 3.5 21.1 1.3 0.6 6.4 2.1 1.5 1.0 0.7 100 merchandise 9 Transport, ware- 1.3 2.7 1.7 2.2 0.9 2.3 2.4 3.0 58.0 2.1 2.0 3.5 10.3 1.2 1.2 1.9 1.1 1.0 0.7 0.6 100 house operatives, packers 10 Personal protection, 0.3 1.0 0.2 1.0 0.1 0.6 0.6 1.4 3.0 80.0 0.8 0.9 4.0 0.6 0.7 2.5 0.5 0.8 0.1 0.9 100 guards and secutrity occupations 11 Hotel, restaurant 3.2 2.1 0.9 1.0 0.6 2.3 6.2 2.9 7.4 1.3 47.5 8.5 6.6 0.5 0.7 3.0 1.2 2.4 1.2 0.8 100 occupation, house- keeping 12 Cleaning, disposal 1.8 3.0 1.5 1.4 0.6 3.3 2.9 1.2 7.2 1.3 5.3 60.9 2.0 0.4 0.5 2.7 0.4 1.9 1.5 0.4 100 occupations 13 Ofﬁce and com- 0.5 0.7 0.4 0.2 0.2 0.5 3.1 5.5 2.6 1.7 1.9 2.1 67.5 1.6 0.7 5.9 1.9 1.4 1.1 0.6 100 mercial services occupations 14 IT and natural sci- 0.7 0.5 1.0 0.5 0.2 0.6 0.8 3.0 1.7 0.8 0.9 0.8 7.3 52.2 4.2 13.2 5.2 0.9 0.5 4.9 100 ence 15 Technical occupa- 0.9 2.0 7.6 1.6 5.9 5.0 2.5 3.6 5.0 1.4 1.7 2.5 8.2 6.2 33.9 6.8 1.9 1.3 0.7 1.5 100 tions Modelling reallocation processes in long-term labour market projections 77 and qualiﬁcation group for the years 2005 to 2011 using Microcensus data. Table 1 shows the aggregate distribu- tion, the so-called ﬂexibility matrix, for the year 2011 for the summarized 20 MOF, where the dark cells highlight the percentage of stayers. Overall, we can see that some groups of persons as distinguished by training OF are more concentrated on fewer exercised OF than others. MOF 20 ‘teaching occupations’ is a classic example of high concen- tration. Next, the elasticity ı of Eq. 4 is retrieved, estimating a model of the log share of stayers on the log wage to reference wage ratio, a constant and an error term. We es- timate this model using the aggregated ﬂexibility matrices over all age, sex and qualiﬁcation groups for the 54 OF cross-sections and the years 2005 to 2011. For more robust results we pool OF of similar qualiﬁcation proﬁles and his- toric wage responsiveness together to estimate this model as four separate ﬁxed-effects panel models. Therefore, in each panel all persons associated with a certain training OF react in the same manner to wages in the model. How- ever, the differences in occupational mobility according to age, sex and qualiﬁcation are accounted for by using the different ﬂexibility matrices for each group in the projec- tion. Panel 1 comprises different OF who have shown high wage responsiveness in the past and consist of high shares of highly educated and very low shares of non-formally qualiﬁed workers. Panel 2 includes highly wage responsive OF with a workforce highly centred in the medium but also in the low qualiﬁcation levels. Panel 3 consists of low wage responsive OF with a similar qualiﬁcation make-up as panel 2. Finally, panel 4 contains miscellaneous OF with historically very low wage responsiveness. Table 2 displays the results of the separate panel regres- sions. Note that we only ﬁnd an elasticity for 36 of the 54 cases. The remaining cases, as for example ‘health-care occupations not requiring a medical practice license’, for which no signiﬁcant elasticity can be found, do not react to wages in the model. In addition, people without any formal qualiﬁcation are assumed to distribute over exercised OF, in which they comprised at least 3% of the workforce in 2011, according to labour demand, while the distribution over exercised OF of those in education are held constant in the projection. It is also likely that the structure of the wage data plays a role in this case. The wage data of gainfully employed persons and the legal censorship in the upper income range probably do not represent an ideal measurement, particularly with regard to the OF of ‘managing directors, auditors, management consultants’ and ‘legal occupations’. In the case of ‘health-care occupations not requiring a medical practice license’, for example, which also show a higher proportion of self- employed persons and a higher income, no positive elasticities can be demonstrated. Nevertheless, because of the absence of a more exact database, it seems appropriate to use the elasticities as given in Table 4 for the baseline projection. Table 1 Occupational ﬂexibility matrix from formal vocational qualiﬁcation to occupation exercised in 2011 for 20 MOF (Continued) MOF formal vocational Switches to MOF exercised (in %) qualiﬁcation 12 34 5 6789 10 11 12 13 14 15 16 17 18 19 20 Total 16 Legal, management 0.4 0.3 0.2 0.2 0.1 0.2 1.3 6.6 1.1 0.8 1.0 0.5 25.1 3.6 0.7 49.3 4.6 0.7 1.0 2.4 100 and business science 17 Media, arts and 0.4 0.6 0.3 0.5 0.5 0.7 2.3 4.0 1.7 0.7 1.8 1.3 9.9 6.0 1.3 7.8 43.9 1.7 2.8 11.9 100 social science 18 Health occupations 0.4 0.7 0.3 0.1 1.3 0.6 3.3 1.6 1.8 0.5 2.0 3.4 6.5 0.5 0.3 1.5 1.0 71.2 2.2 0.9 100 19 Social occupations 0.4 0.5 0.2 0.1 0.1 0.3 1.6 1.0 1.2 0.4 1.6 2.7 5.0 0.5 0.1 2.4 1.7 3.9 66.4 10.0 100 Teaching occupa- 0.2 0.2 0.2 0.2 0.1 0.3 0.8 1.3 1.1 0.2 1.2 1.8 3.9 0.9 0.2 1.5 2.5 1.7 3.4 78.4 100 tions Without formal 3.0 7.8 3.7 4.3 1.7 5.6 7.0 3.1 14.8 1.9 9.9 17.5 7.3 1.2 1.0 2.0 2.2 3.5 1.8 0.8 100 vocational qualiﬁca- tion Still in training 2.0 0.8 6.8 3.9 3.3 4.2 7.2 7.0 6.0 1.5 8.6 1.9 15.6 3.7 2.2 1.7 5.2 10.8 4.2 3.5 100 Source: German Mikrocensus 2011, own calculations (BIBB) 78 T. Maier et al. Table 2 Wage elasticitiy of stayers ı by OF (2005–2011) OF ı Panel 1: 2.2 21 Engineers | 22 Chemists, physicists, scientists | 31 Advertising specialists | 36 Administrative occupations in the public industry | 51 Journalists, librarians, translators, related academic research occupations | 46 Designers, photographers, advertising creators | 24 Technical draughtsmen/draughtswomen, related occupations Panel 2: 2.59 16 Cooks | 34 Packers, warehouse operatives, transport processors | 40 Auxiliary ofﬁce occupations, telephone operators | 52 Body care occupations Panel 3: 1.27 1 Agriculture, husbandry, forestry, horticulture | 2 Miners and mineral extraction workers | 5 Paper manufacture, paper processing, printing | 9 Vehicle and aircraft construction, maintenance occupations | 10 Precision engineering and related occupations | 14 Bakers, pastry cooks, production of confectionary goods | 15 Butchers | 18 Construction occupations, wood and plastics manufacture and processing occupations | 41 Personal protection, guards| 49 Social occupations | 54 Cleaning and disposal occupations Panel 4: 0.57 6 Metal production and processing | 7 Metal, plant, and sheet metal construction, installation, ﬁtters | 13 Textile processing, leather man- ufacture | 17 Production of beverages, foods and tobacco, other nutrition occupations | 23 Technicians | 25 Surveying and mapping | 26 Specialist skilled technicians | 27 Sales occupations (retail) | 30 Other commercial occupations (not including wholesale, retail, bank- ing) | 32 Transport occupations | 35 Managing directors, auditors, management consultants | 39 Commercial ofﬁce occupations | 44 Legal occupations | 53 Hotel and restaurant occupations, housekeeping Source: German Mikrocensus and EHD from 2005 until 2011; own calculations Note that the result that workers and employers of dif- changes. In the QuBe model, mainly in favour for keeping ferent training occupations and different economic sectors, the model simple such that results are more transparent, respectively, do not adjust to changes in the labour market in this, however, is not accounted for. the same magnitude, conforms to the discussion of Sect. 2: The reallocation process is also subject to inﬂuences other 5 Scenario comparisons than wages. These are (only) implicitly contained in the QuBe model. However, the comparison of these wage elasticities to In this section, we will display the magnitude of effect results of other studies is limited. The reasons are that (a) of the previously described reallocation mechanism of the these elasticities do not resemble causal effects, but also QuBe model on the projection outcomes and the practi- capture other effects which relate to wages and mobility; cal recommendations based on them. For this purpose, we and (b) because they are based on the relation of the oc- estimate labour demand and supply for various scenarios cupation speciﬁc reference wage with the stayer rate (see concerning a different occupational ﬂexibility behaviour or Eq. 4). Because the reference wage contains also the own wage setting assumptions. Firstly, in Sect. 5.1 we demon- wage of each occupation proportional to the historic ﬂex- strate the overall effect on the projection results from con- ibility, this relation is higher than only looking at outside sidering versus not considering a reallocation process. Sec- wages. Therefore, these elasticities are relatively high. ondly, in Sect. 5.2 we show, which effect can be attributed Further, these wage elasticities of the stayer rate are kept to the dynamics of worker adjustments with respect to constant over the projection period. Departing from this wages. After this, we continue with scenario comparisons assumption would potentially also relevantly affect the pro- to highlight the limitations to wage adjustments in resolv- jection outcomes. It is plausible, for example, that tech- ing labour shortages in the QuBe model and by that the nological progress has an impact on the extent to which importance of other determinants for occupational mobility, wages drive mobility decisions. New technologies are sug- which are only implicitly modelled. We show that these lim- gested to lead to either an increase of complexity of tasks to itations have a meaningful impact on the deduction of rec- be performed by workers or a ‘deskilling’ of tasks, where ommended actions to alleviate occupation-speciﬁc labour specialized skills become redundant (cf. Ben-Ner and Ur- shortages. For this purpose, thirdly, in Sect. 5.3 we show tasun 2013). A change in the skill requirements may lead how the economic environment of the employer matters for to a change in mobility behaviour following the reason- the result of different wage setting policies and the feasibil- ing of Geel and Backes-Gellner (2011) and Gathmann and ity of such wage scenarios according the the QuBe model. Schönberg (2010). Different outside opportunities may then Lastly, in Sect. 5.4 we complement the previous result by also translate into a different receptiveness for relative wage deriving the optimal stayer rates for the occupations and K Modelling reallocation processes in long-term labour market projections 79 rd Fig. 2 Skill shortages and surpluses with and without reallocation in 2005–2030. Source: QuBe project 3 wave; own calculations discuss to what extent these stayer rates are achievable by MOF 18 ‘Health occupations’. The technicians are frequent the means of wage policies. Note that throughout the fol- movers with a stayer share of only 33.9% (cf. Table 2)and lowing section, we implement the scenario assumptions on are able to ﬁnd work in a lot of different MOF. Also, the the level of the 54 OF. However, for a better visualization supply of skilled technicians is decreasing strongly until the results are always presented on the level of the 20 MOF. 2030 (see the decreasing surplus in the left hand graph over time) due to demographic change and retirement of 5.1 Implementing occupational ﬂexibility the so-called ‘baby boomers’, who are more often trained in a manufacturing or technical occupation than younger To start with, Fig. 2 illustrates the effect of implement- cohorts. ing a reallocation process by comparing the projection re- The health occupations, however, face another problem: sults of the QuBe baseline scenario (right hand side) with Workers in this ﬁeld are to a great extent loyal to their a scenario, in which workers were not allowed to switch occupation as indicated by their stayer rate of 71.2% (cf. and employers could not substitute skilled for unskilled or Table 2). Here as well, not enough workers are being trained workers from different disciplines (left hand side). In the in this ﬁeld (again note left hand graph), while the demands latter scenario, the projection results suggest that vast labour are increasing due to the ageing of the population (Maier shortages are possible in 9 out of the 20 MOF. According and Afentakis 2013). to this, for 8 of these MOF shortages should have actu- Ultimately, the total deﬁcit in the baseline scenario is ally already been visible in 2010. In 2030 the deﬁcit would 0.3 million workers only, thus, revealing the substantial grow to about 4.9 million skilled workers in this scenario. In impact of a reallocation mechanism on the projection re- comparison, taking the reallocation mechanism into account sults. Therefore, not taking the empirically veriﬁable oc- balances the labour market in all but 4 of these occupations; cupational mobility into account at all would exaggerate however, shortages appear until 2030 in 5 additional MOF. possible future shortages. Interestingly, now shortages could become especially im- minent in the MOF 15 ‘Technical occupations’ and the K 80 T. Maier et al. in the MOF 5 ‘Other processing, producing and maintain- ing occupations’ (HHI = 0.12). This MOF contains, for example, the textile processors, which have to switch occu- pations more often as the textile industry in Germany is be- ing downsized. Only persons currently in education (HHI = 0.08) and persons with no vocational training (HHI = 0.09) were more evenly distributed. We found the highest concen- tration in the MOF 10 ‘Personal protection, guards and se- curity occupations’ (HHI = 0.64). Also the MOF 18 ‘Health occupations’ (HHI = 0.52) and the MOF 20 ‘Teaching oc- cupations’ (HHI = 0.62) are highly concentrated. These 3 MOF have also the highest stayer rates. The mean HHI equals 0.32 weighted by the labour force participants in each training MOF. In Fig. 3, we now contrast the difference between con- Fig. 3 Differences in HHI due to structural change (2030–2011) and wage development (‘no wage response’ vs. ‘baseline’). Source: QuBe stant and wage responsive ﬂexibility. On the vertical axis, rd project 3 wave; own calculations we plot the pure time trend of the HHI in the 20 MOF, i. e. the HHI in 2011 compared to 2030 of the ‘no wage 5.2 Implementing ﬂexibility dynamics response’ scenario. On the horizontal axis, we plot the HHI differences in 2030 between the baseline scenario Next, we will further analyse how the wage dynamics of with wage elastic ﬂexibility and that without. Note how occupational mobility as implemented in the baseline sce- shifts along the vertical axis visualize pure structural ef- nario of the QuBe model impact the projection results. For fects, while shifts along the horizontal line show how the this purpose, consider a world, in which workers did not concentration of the workforce on exercised occupations respond to wage changes, even if they occurred in occu- increases or decreases as a result of wage incentives. pations in which they could have very likely also found Fig. 3 illustrates that concentration hardly changes over work and proﬁted from a wage gain. Thus, in such a world time due to changes in the labour force composition. An the probability to stay and switch are time invariant. How- exception is the MOF 10 ‘Personal protection, guards and ever, note that aggregate mobility in the occupations does security occupations’. This MOF interestingly has the high- change over time, as the age and qualiﬁcation composi- est HHI in 2011, which, however, is decreased by almost tion of the workforce changes due to demographic change. –0.05 units due to structural change only. Note that the Therefore, comparing projection results for such a world other outlier of MOF 2 ‘Auxiliary workers, janitors’ is ac- with the QuBe baseline scenario enables us to disentangle tually very small in terms of trained labour supply. The the effect of wage responsiveness from structural effects. wage mechanism of the baseline scenario leads to a higher To visualize the concentration of the workforce, i. e. degree of dispersion over exercised MOF in most training the possibilities to work with a certain formal vocational MOF. Wage responses cause the highest reduction in con- qualiﬁcation in different OF, we calculate the Herﬁndahl- centration in the MOF 12 ‘Cleaning, disposal occupations’ Hirschman-Index (HHI henceforth; cf. Hirschmann 1964) and the MOF 18 ‘Health occupations’. Here, the projected for the 20 MOF. wage growth fails that of alternative occupations in other MOF leading to higher occupational switching and, there- fore, a greater dispersion. Note that the observed effect on HHI = (5) to P the MOF 18 can be purely attributed to a change in disper- o=1 o=1 sion of body care occupations, as doctors and nursing staff where x represents the amount of workers in the exercised do not dynamically respond to wages in the baseline sce- MOF o with the training MOF to for which the HHI is nario (cf. Table 2). Also note that the MOF 12 and MOF 18 evaluated. As there are 20 MOF over which the labour force still have some of the highest stayer shares in 2030. In participants of a training occupation can disperse, HHI 2 contrast, in the MOF 16 ‘Legal, management, and business Œ1=20; 1, where the minimum value of 0.05 indicates an science occupations’ or 19 ‘Social occupations’ the wage even distribution over all exercised MOF and the maximum related increase in concentration level out the dispersion value of 1 indicates perfect concentration on the training due to structural effects, such that these occupations have occupation. almost stable HHIs over time. For the year 2011, the ﬂattest empirical distribution is The resulting labour demand and supply for each sce- observed for persons with a formal vocational qualiﬁcation nario in 2030 can be retrieved from Table 3. It can be K Modelling reallocation processes in long-term labour market projections 81 Table 3 Labour demand and supply in 1000 persons by 20 MOF in 2030 in the baseline and ‘no wage response’ scenario Main Occupational Field Baseline ‘No wage response’ (MOF) Supply Demand Diff Supply Demand Diff 1 Raw material processing occupations 907.6 898.6 8.9 929.8 899 30.8 2 Auxiliary workers, janitors 1088.3 1103.6 –15.3 1073 1103.4 –30.5 3 Metal production and processing, installation, 1603.9 1576.5 27.3 1601.7 1575.9 25.8 electrical occupations 4 Construction, woodworking, plastic manufac- 1442.2 1451.1 –8.9 1466.8 1450.6 16.2 ture and processing occupations 5 Other processing, producing and maintaining 848.4 831.9 16.4 858.2 831.7 26.4 occupations 6 Machinery and equipment steering and main- 1772.2 1675.1 96.9 1890 1674.8 215 tainance occupations 7 Commodity trade in retail 2056.5 2059.7 –3.2 1968.5 2058.1 –89.5 8 Commodity trade merchandise 2154.4 2024.3 130.1 2213.5 2023.8 189.7 9 Transport, warehouse operatives, packers 3165.1 3194.6 –29.5 3038.8 3194.2 –155.2 10 Personal protection, guards and security occu- 677.5 657.6 19.9 681.3 657.6 23.7 pations 11 Hotel, restaurant occupation, housekeeping 2158.3 2176 –17.6 2079.3 2170.5 –91.1 12 Cleaning, disposal occupations 2051.7 2052.8 –1.1 2004.4 2052.6 –48.3 13 Ofﬁce and commercial services occupations 6301.3 5545.5 755.8 6479.6 5548.7 931 14 IT and natural science 2369.6 2214 155.6 2390.5 2213.7 176.7 15 Technical occupations 1188.7 1249.2 –60.6 1152.6 1248.9 –96.4 16 Legal, management and business science 2850.8 2679.1 171.7 2811.6 2678.9 132.6 17 Media, arts and social science 1747.9 1792.9 –44.9 1740 1792.2 –52.1 18 Health occupations 3863.7 4016.6 –152.9 3847.9 4027.3 –179.5 19 Social occupations 1739.5 1662 77.4 1737 1662.6 74.4 20 Teaching occupations 1790.6 1500.3 290.4 1813.8 1501.6 312.2 rd Source: QuBe project 3 wave; own calculations observed that without accounting for wage responsive ﬂex- 5.3 The limitations to wage adjustments ibility behaviour, the total deﬁcit equals about 740,000 per- sons. This is more than twice the deﬁcit of the baseline We now examine the impact of wage policies in greater scenario with dynamics, which amounts to only 340,000 detail and point out the importance of their limitations in persons. Thus, 400,000 workers, which would be unem- the QuBe model for the interpretation of results. Shortages ployed in other surplus occupations in the projection, are are partly projected, due to inferior wage developments in redistributed to the shortage occupations where wages are these occupations. Because outside wage opportunities are rising in the baseline scenario. growing more strongly than in the own training OF, work- However, in the MOF 4 ‘Construction, woodworking, ers – where empirically veriﬁable – more often decide to plastic manufacture and processing occupations’ the labour switch occupations. Employers can take advantage of this market actually gets tighter due to wage dynamics. Here, by raising wages in occupations where labour is scarce. although the wage responsiveness of ﬂexibility is actually However, they are (depending on the industry) constraint not too high, the projected development of the outside wage by price competition with ﬁrms abroad and consumer de- options induces the workforce to switch more often to other mand. This is reﬂected in the QuBe model. To show to what occupations. In this case, the possibility of a future labour extent employers can strategically use wage adjustments in shortage may be understated when dynamic behaviour in this model, we implement further wage increases for short- occupational ﬂexibility is not accounted for. Ultimately, we age occupations (as singled out by the baseline projection can conclude that assumptions about the wage responsive- results). We consider a scenario, where wage growth in the ness of labour mobility are crucial for assessing possible shortage occupations is increased by 10% until 2030 com- future labour market outcomes. pared to the baseline wage development. Note that this rep- resents an increase of a little more than 0.5% every year un- til 2030 on top of the projected wage growth of the baseline scenario. Since this represents a relatively small change, in K 82 T. Maier et al. Table 4 Labour demand and supply in 1000 persons by 20 MOF in 2030 in baseline model and different wage scenarios Main Baseline ‘10%-increase’ ‘20%-increase’ Oc- Supply Demand Diff Supply Demand Diff Supply Demand Diff cu- 1 Raw material processing occu- 907.6 898.6 8.9 906 895.6 10.4 904.5 892.6 11.9 pa- pations tional 2 Auxiliary workers, janitors 1088.3 1103.6 –15.3 1090 1097.8 –7.7 1091.6 1092.4 –0.8 Field (MOF) 3 Metal production and process- 1603.9 1576.5 27.3 1602.1 1564.7 37.4 1600.1 1553.4 46.7 ing, installation, electrical occu- pations 4 Construction, woodworking, 1442.2 1451.1 –8.9 1448.4 1433.8 14.6 1453.9 1417.6 36.3 plastic manufacture and pro- cessing occupations 5 Other processing, producing 848.4 831.9 16.4 843.5 826.5 17.1 839.2 821.3 17.9 and maintaining occupations 6 Machinery and equipment 1772.2 1675.1 96.9 1767.5 1663.3 104.2 1763 1652 110.9 steering and maintainance oc- cupations 7 Commodity trade in retail 2056.5 2059.7 –3.2 2047.1 2029.9 17.2 2037.5 2001.9 35.6 8 Commodity trade merchandise 2154.4 2024.3 130.1 2152.2 2010.2 142 2150.2 1996.6 153.6 9 Transport, warehouse opera- 3165.1 3194.6 –29.5 3170.2 3176.3 –6.2 3174.1 3159.3 14.8 tives, packers 10 Personal protection, guards and 677.5 657.6 19.9 681.5 660.1 21.4 685.3 662.2 23 secutrity occupations 11 Hotel, restaurant occupation, 2158.3 2176 –17.6 2111.9 2094.7 17.2 2071.4 2024.6 46.8 housekeeping 12 Cleaning, disposal occupations 2051.7 2052.8 –1.1 2045.6 2029.1 16.6 2038.9 2007 31.9 13 Ofﬁce and commercial services 6301.3 5545.5 755.8 6296.2 5538.6 757.6 6291.8 5531.5 760.3 occupations 14 IT and natural science 2369.6 2214 155.6 2361.4 2207.1 154.3 2353.4 2200.4 153 15 Technical occupations 1188.7 1249.2 –60.6 1202.6 1243.1 –40.4 1214.1 1237.2 –23.1 16 Legal, management and busi- 2850.8 2679.1 171.7 2849.6 2667.8 181.8 2848.8 2657.1 191.7 ness science 17 Media, arts and social science 1747.9 1792.9 –44.9 1766.1 1789.2 –23.1 1782.7 1785.2 –2.5 18 Health occupations 3863.7 4016.6 –152.9 3899.6 4017 –117.4 3933.5 4016.4 –83 19 Social occupations 1739.5 1662 77.4 1740.3 1673 67.3 1741.9 1683.2 58.7 20 Teaching occupations 1790.6 1500.3 290.4 1796.3 1513.6 282.7 1802.2 1526 276.2 rd Source: QuBe project 3 wave; own calculations a second scenario we increase wage growth in the shortage by even 225,000 persons compared to the baseline scenario. occupations by 20%, i. e. an additional increase of a lit- However, the labour market is balanced in only one addi- tle more than 1% every year until 2030. The results are tional MOF compared to a 10% increase until 2030, namely presented in Table 4. the MOF 9 ‘Transport, warehouse operatives, packers’. We The results show (cf. Table 4) that with a wage increase can see that the balance in these MOF is mainly achieved by of an additional 10% until 2030 for shortage occupations, a reduction in labour demand. Since labour productivity re- labour shortages will be reduced by about 140,000 per- mains unchanged, note that this corresponds to a reduction sons in 2030, so that the total deﬁcit in this scenario equals in production or service provision, respectively. In these 195,000 persons. Shortages could be prevented in 4 of the occupations, outside price pressures are too high, such that 9 shortage MOF of the baseline scenario, namely in the large wage adjustments are infeasible for employers without MOF 4 ‘Construction, woodworking, plastic manufacture reducing their output. Here, it is more realistic that alterna- and processing ocupations, MOF 7 ‘Commodity trade in tive strategies would be used to retain workers or workers retail’, MOF 11 ‘Hotel, restaurant occupation, housekeep- would be hired from abroad to keep the wage level low. ing’, and the MOF 12 ‘Cleaning, disposal occupations’. The other shortage MOF, for which a shortage is pro- Looking at the results, of the 20% increase in wage growth jected until 2030 even after an additional wage increase of for baseline shortage occupations, the total deﬁcit of labour 20%, are the MOF 2 ‘Auxiliary workers, janitors’, MOF 15 supply equals about 115,000 persons, which is a reduction ‘Technical occupations’, MOF 17 ‘Media, arts and social K Modelling reallocation processes in long-term labour market projections 83 of practical advice based on calculations of a projection model. 5.4 The ‘optimal’ ﬂexibility In the following, we examine, what kind of adjustments in the occupational ﬂexibility behaviour would be needed to distribute unemployed workers evenly and to overcome labour shortages in every OF in 2030. Thus, this scenario entails a redistribution of the labour supply from surplus to shortage occupations. Also looking at the results from the previous section, we assess how wages can serve to achieve the resulting differences in the stayer rate according to the assumptions of the baseline scenario. Technically, we apply a RAS procedure. The RAS al- Fig. 4 Needed adjustments of occupational ﬂexibility to achieve rd equal unemployment rates in 2030. Source: QuBe project 3 wave; gorithm (cf. Bacharach 1970; Leontief 1951) is an itera- own calculations tive method of biproportional ﬁtting of matrices, which is used to estimate elements of an unknown matrix based on science occupations’, and MOF 18 ‘Health occupations’. known row and column sums and an initial estimate of the In all of these MOF, demand remains relatively stable, sug- matrix. Transferred to this exercise, the RAS algorithm ﬁts gesting that here price pressures are less dominating, be- the cells of the ﬂexibility matrix of 2030, such that column cause production or service provision cannot simply be re- totals, i. e. labour supply in the exercised OF, are such that duced. We leave the MOF 2 out of the discussion as they in every OF an equal unemployment rate is achieved. In comprise a very small group of people and are not asso- doing so, the algorithm loops over occupations – starting ciated with dynamic behaviour in the QuBe model mainly with that with the highest unemployment rate – and redis- due to data restrictions. The MOF 15 and 17 have com- tributes the difference between the baseline surplus supply parably lower stayer rates of 39% and 43%, respectively, and that needed to achieve the targeted unemployment rate because their labour can be applied in very diverse ﬁelds. to other occupations. The reallocation is proportional to the Here, the deﬁcit is more severe in the MOF 15, mainly be- initial ﬂexibility matrix of the baseline scenario, such that cause workers trained in this ﬁeld react much less to wage workers trained in surplus occupations switch more (have impulses. Most of the occupations in MOF 17 are attached a smaller stayer rate); however, to the same extent into the with a wage elasticity of 2.2 in the baseline scenario, in- same exercised occupations. dicating that career seeking is a major determinant in the Fig. 4 visualizes the change in ﬂexibility again using dif- occupational ﬂexibility behaviour of journalists, designers ferences in the HHI indicating growing or declining concen- etc. In contrast, most of the occupations in the MOF 15 tration in the MOF. Here, the difference in the HHI between only react to wages with an elasticity of 0.57, suggesting 2011 and 2030 in the baseline projection is plotted on the that here other factors as for example better working con- vertical axis against the HHI difference between the 2030 ditions may strongly inﬂuence mobility decisions. workforce of the baseline projection and the scenario using In the MOF 18 it is only the body care occupations, the optimal occupational ﬂexibility matrix on the horizon- which react to wage impulses. Increasing wages cannot tal axis. MOF plotted to the left (right) of the 0 benchmark reduce the shortage of doctors and nursing staff, because on the horizontal axis, indicate a need for a higher (lower) the baseline QuBe projection reﬂects that their fairly high ﬂexibility as compared to the baseline assumptions in order occupational loyalty is not signiﬁcantly driven by wage in- to clear the labour market in 2030. centives. Also, a wage increase in these occupations does Overall, the majority of the MOF actually should be not considerably raise the inﬂow of labour supply into these more ﬂexible in order to correspond optimally to labour occupations from other ﬁelds, which simulates the effect of demand. Especially, persons in the MOF 20 ‘teaching oc- strong working regulations concerning qualifying creden- cupations’ but also the MOF 13 ‘ofﬁce and commercial ser- tials and approbations (see also Pollmann-Schult 2006). vices occupations’, for which vast surpluses are projected Overall, we can conclude here that accounting for the due to demographic change and a rising educational at- limitations to wage setting policies within the projection tainment in these occupations, should more often consider model has signiﬁcant impacts on the feasibility of scenarios switching their occupation in the future. In the MOF 20, aimed at overcoming shortages. This has important conse- the share of stayers would have to be reduced from 79.4 quences for policy consulting and enhances the credibility to 66.6% in 2030. In the MOF 13 a reduction of the share K 84 T. Maier et al. of stayers to 61.6% from its level of 67.2% in the baseline most often is the ultimate aim of long-term labour market projection in 2030 would be needed. Note that this MOF projection. also contains the public administrates, which mainly drive this result, here. They alone would need a reduction of the stayer ratio by more than 12 percentage points. However, 6 Conclusion and discussion in both of these MOF, workers do not react to increases in outside wages in the QuBe model and are very loyal to In this paper, we discuss and illustrate the necessity of im- their training occupations (cf. Table 1 and 2). This poses plementing a dynamic reallocation process of labour sup- a challenge of achieving such a reduction in stayer rates. ply into labour market projections and how the underlying Likely, this could not be accomplished by increasing wages assumptions strongly inﬂuence the plausibility of the pro- in related ﬁelds, as other underlying factors as for example jection results and their interpretation for policy consulting. work place stability or reconciliation of family and work Long-term projections have become very popular for guid- are stronger motivators for high stayer rates in these occu- ance in political decision-making. Therefore, it is essential pations. that the model set-up reﬂects country-speciﬁcs and can draw In contrast, persons trained in the MOF 18 ‘Health oc- a plausible image of the possible future developments. In cupations’ would need to stay in their occupation more Germany, therefore, it is essential that a projection model often. The projected stayer rate of 67.8% in 2030 in the (a) represents the occupational dimension of the German baseline scenario would have to increase to 71.9%. This labour market and (b) reﬂects the extent to which workers complements the results of the previous section: Because skilled in different occupations can be substituted for each switches into these occupations are quite unlikely due to other (Helmrich and Zika 2010). These two aspects are es- work regulations, the needed increase in stayers would only sential for an assessment of possible reallocations of labour be achievable via an even greater occupational loyalty or in- supply in respond to imminent shortages. creased training of new supply. Since outside wages are not The BIBB-IAB qualiﬁcation and occupational ﬁeld pro- signiﬁcantly important to doctors and nursing staff, the re- jections (Maier et al. 2014) is to our knowledge the only sults again stress the impact of other factors, as for example long-term projection model, which explicitly formulate working conditions, on making these occupations more at- such a reallocation process. In this model, the central link tractive for policy recommendations to realize the increase between demand and supply is wages: Employers raise in labour supply. wages in shortage occupations to make work in these ﬁelds Interestingly the shortage MOF 15 ‘Technicians’, would more attractive and workers react to relative changes in their hardly need any ﬂexibility adjustments at all according outside wage opportunities and adjust their intent to stay. to this calculation. Their optimal ﬂexibility would entail The great degree of detail of the model by 63 economic a stayer share of 35.7% in 2030. Therefore, the adjustment sectors and 54 occupational ﬁelds provides a thorough de- from its baseline value of 33.2% would amount to merely scription of the diverse adjustment behaviours of different 2.5 percentage points. Here, the redistribution from surplus groups of market participants. In this way, the projection re- ﬁelds is high enough such that only a small adjustment in sults also implicitly capture reallocation behaviours, which the stayer rate sufﬁces to balance the submarket for techni- are not driven by wage and scarcity, respectively. cians. We ﬁnd that almost 70% of the additional workforce Our results show that not accounting for occupational in this MOF would be recruited from outside (mainly engi- ﬂexibility at all, i. e. not modelling any reallocations in the neers and electrical occupations). Here, wage policies may labour market, would project vast shortages of almost 5 mil- serve to attract workers from related ﬁelds to some extent, lion skilled workers in 9 of the 20 main occupational ﬁelds however the persisting shortages even after large wage in- in 2030. Compared to this, the baseline scenario, which ac- creases (cf. Sect. 5.3) suggest that again working conditions counts for dynamic adjustments on both sides, would only in this ﬁeld may be more promising to target. project a total deﬁcit of about 340,000 workers in 2030. In summary, for an optimal distribution of unemployed However, the reallocation process can be directly linked to workers over the exercised occupations, stayer rates for shortages, which now appear in ‘health occupations’ and many training occupation would have to differ. As already ‘technical occupations’. In both of these main occupational discussed in the previous section, wages are often an infea- ﬁelds inﬂows from other ﬁelds would not sufﬁce to balance sible tool to reach the optimum, here. In the QuBe model, out the outﬂows of skilled workers to other related ﬁelds. alternative determinants for occupational mobility are im- Next, looking at the effect of dynamic adjustments of portant for the interpretation of the results, although they the ﬂexibility behaviour of workers, we compare the base- are only implicitly accounted for. In the end, this is essential line scenario to a scenario, where shares of stayers do not for deriving recommendations for practical actions, which respond to wages. We ﬁnd that dynamics can account for a difference in the deﬁcit of labour supply of about 400,000 K Modelling reallocation processes in long-term labour market projections 85 people in 2030. Shortages in the ‘Construction, woodwork- ups where for example dynamics evolve subject to techno- ing, plastics manufacture and processing occupations’ ac- logical progress are possible and maybe a fruitful ﬁeld for tually become more severe in the projection results after research. However, when advancing model set-ups in these considering a dynamic adjustment of workers. Here, wage ways, the transparency of results has to always be kept in dynamics reﬂect the tension between price and employer mind as well (c.f. Wilson 2001). competition for labour supply. Lastly, in the discussed model, the potential of the of- Furthermore, we illustrate how also the limitations to fered amount of hours by the labour force has been assumed wage dynamic adjustments as captured by the QuBe model to be stable during the projection period (Zika et al. 2012). inﬂuence the interpretation of results and the derived rec- Furthermore, it is assumed that participation rates follow ommendations for practical actions. For this, we look at the an increasing trend and migration inﬂow to Germany is th effect of different wage policies. We compare a 10% and kept constant according to the 12 Coordinated Population a 20% increase of wages until 2030 for shortage occupa- Forecast of the Federal Statistical Ofﬁce. Of course, these tions. We see that in the QuBe model these wage increases measures could in principal also work as dynamic mecha- would be able to balance some occupational submarkets, nisms in long-term labour market projections. In fact, this however, mainly by a reduction of labour demand and, thus, may work better for employers in occupations with strong a lower production or service provision in the economy. For wage setting constraints and workers in occupations with the remaining shortage occupations in these scenarios, we low wage responsiveness. As this has not been done thus discuss how wages as a policy tool simply are not effec- far, in future studies it would be very interesting to assess tive given the QuBe assumptions about wage dynamics of the differences in projection results and policy advice be- occupational mobility. Especially for technicians, doctors, tween obtained from projections using these different mech- and nursing staff other factors related to working conditions anisms. may be more important for political actions. In the case of Open Access This article is distributed under the terms of the health occupations, also working regulations play an im- Creative Commons Attribution 4.0 International License (http:// creativecommons.org/licenses/by/4.0/), which permits unrestricted portant role, which limits the extent to which workers from use, distribution, and reproduction in any medium, provided you give outside can be recruited for this ﬁeld. appropriate credit to the original author(s) and the source, provide a We complement these results further, by calculating the link to the Creative Commons license, and indicate if changes were ‘optimal’ ﬂexibility of the workforce, which would evenly made. distribute unemployed workers over the occupations. We ﬁnd that most of the workforce would have to be more ﬂexible. In contrast, health personnel would need to stay more often within their training occupation. As they do not respond to wages empirically, again working conditions but also increased training of new supply may be more feasible policy implementations. Surprisingly, in the case of techni- cians no large adjustment of mobility behaviour would be needed, because also an increased inﬂow of workers from related ﬁelds would help to balance out deﬁcits of labour supply. In this ﬁeld, the sufﬁcient provision of labour sup- ply may be achieved, both to their own extent, by increasing wages and improving work conditions, but also by provid- ing persons with related educational backgrounds further educational training to enhance speciﬁc needed skills. The results illustrate how for the derivation of plausible policy recommendation also the limitations to reallocations are central to modelling. Based on the QuBe model, how- ever, we can only discuss the relative importance of other driving factors of occupational mobility in light of the re- strictions of the wage dynamics. Therefore, also integrating, for example, working conditions into long-term labour mar- ket projection models may be an intriguing ﬁeld of further studies. Furthermore, throughout our analyses we assume that the response of workers to outside wages in their mo- bility decisions is time invariant. Here, as well different set- K 86 T. Maier et al. Appendix Table 5 Major Occupational MOF OF Fields (MOF) and Occupational 1 Raw material processing 1 Agriculture, husbandry, forestry, horticulture Fields (OF) occupations 2 Miners and mineral extraction workers 2 Auxiliary workers, janitors 20 Auxiliary workers without further speciﬁed task 42 Janitors 3 Metal production and 7 Metal, plant and sheet metal construction, installation, processing, installation, ﬁtters electrical occupations 11 Electrical occupations 4 Construction, woodworking, 18 Construction, woodworking, plastics manufacture and plastic manufacture and processing occupations processing occupations 5 Other processing, producing 3 Stoneworking, construction materials production, ceram- and maintaining occupations ics and glass related occupations 9 Vehicle and aircraft construction, maintenance occupa- tions 10 Precision engineering and related occupations 13 Textile processing, leather manufacture 15 Butchers 6 Machinery and equipment 4 Chemical and plastics occupations steering and maintainance 5 Paper manufacture, paper processing, printing occupations 6 Metal production and processing 8 Industrial mechanics, tool mechanics 12 Weaving occupations, textile manufacturers, textile ﬁnishers 17 Production of beverages, food and tobacco, other nutri- tion occupations 7 Commodity trade in retail 27 Commodity trade in retail 8 Commodity trade 28 Wholesale/retail service occupations merchandise 30 Other commercial occupations (not including wholesale, retail, banking) 9 Transport, warehouse 19 Goods inspectors, dispatch, processing operators operatives, packers 32 Transport and logistics occupations 33 Aviation, shipping occupations 34 Packers, warehouse and transport occupations 10 Personal protection, guards 41 Personal protection, guards and secutrity occupations 43 Security occupations 11 Hotel, restaurant occupation, 14 Bakers, pastry cooks, production of confectionary goods housekeeping 16 Cooks 53 Hotel and restaurant occupations, housekeeping 12 Cleaning, disposal occupa- 54 Cleaning and disposal occupations tions 13 Ofﬁce and commercial 29 Banking and insurance professionals services occupations 36 Administrative occupations in the public sector 37 Finance, accounting, bookkeeping 39 Commercial ofﬁce occupations 40 Auxiliary ofﬁce occupations, telephone operators 14 IT and natural science 21 Engineers 22 Chemists, physicists, scientists 38 Core IT occupations K Modelling reallocation processes in long-term labour market projections 87 Table 5 Major Occupational MOF OF Fields (MOF) and Occupational 15 Technical occupations 23 Technicians Fields (OF) (Continued) 24 Technical draughtsmen/draughtswomen, related occupa- tions 25 Surveying and mapping 26 Specialist skilled technicians 16 Legal, management and 35 Managing directors, auditors, management consultants business science 44 Legal occupations 17 Media, arts and social 31 Advertising specialists science 45 Artists, musicians 46 Designers, photographers, advertising creators 51 Journalists, librarians, translators, related academic research occupations 18 Health occupations 47 Healthcare occupations requiring a medical practice licence 48 Healthcare occupations not requiring a medical practice licence 52 Body care occupations 19 Social occupations 49 Social occupations 20 Teaching occupations 50 Teaching occupations Table 6 Structure of the NACE Rev. 2 Classiﬁcation of Economic Activities used in the Projection Divisions of the economic sectors (collated) 1 Agriculture 2Forestry 3 Fishing 4 Mining, extraction of stones and earth 5 Manufacture of food and drink, tobacco processing 6 Manufacture of textiles, clothing, leather goods and shoes 7 Manufacture of wood, wicker, basket and cork goods (not including furniture) 8 Manufacture of paper, cardboard and of paper and cardboard products 9 Manufacture of printing products, reproduction of sound, picture and data storage media 10 Manufacture of coke and reﬁned petroleum products 11 Manufacture of chemical products 12 Manufacture of pharmaceutical products 13 Manufacture of rubber and plastic products 14 Manufacture of glass products, manufacture of ceramics, processing of stones and earth 15 Metal production and processing 16 Manufacture of metal products 17 Manufacture of computer, electronic and optical products 18 Manufacture of electrical equipment 19 Engineering 20 Manufacture of motor vehicles and motor vehicle components 21 Other vehicle construction 22 Manufacture of furniture and other goods 23 Repair and installation of machines and equipment 24 Energy supply 25 Water supply 26 Sewage, waste disposal, materials recovery 27 Construction sector 28 Motor vehicle trade, maintenance and repair of motor vehicles K 88 T. Maier et al. Table 6 Structure of the NACE Rev. 2 Classiﬁcation of Economic Activities used in the Projection (Continued) Divisions of the economic sectors (collated) 29 Wholesale (not including the motor vehicle trade) 30 Retail (not including retail of motor vehicles) 31 Land transport and transport in pipelines 32 Shipping 33 Aviation 34 Warehousing, other transport service providers 35 Post, courier and express services 36 Hotel and restaurant trade 37 Publishing 38 Audiovisual media and radio 39 Telecommunications 40 IT and information service providers 41 Financial services providers 42 Insurance and pension funding 43 Activities associated with ﬁnancial and insurance services 44 Real estate 45 Legal and tax consultancy, management consultancy 46 Architectural and engineering companies, technical support 47 Research and development 48 Advertising and market research 49 Freelance, scientiﬁc, technical services (not mentioned elsewhere), veterinary medicine 50 Renting of mobile goods 51 Placement and hiring of workers 52 Travel agencies and tour operators 53 Service providers (not mentioned elsewhere) 54 Public administration, defence, social security 55 Education and teaching 56 Healthcare system 57 Residential homes and social services 58 Art and culture, gambling 59 Sport, entertainment and recreation 60 Lobbying, religious associations 61 Repair of computers and used goods 62 Other providers of mainly personal services 63 Housekeeping services Gesellschaft: Problemlagen und betriebliche Reaktionen. IAB References Forschungsbericht 13/2012. (2012) Ben-Ner, A., Urtasun, A.: Computerization and skill bifurcation: the Abraham, M., Damelang, A., Schulz, F.: Wie strukturieren Berufe role of task complexity in creating skill gains and losses. Ind La- Arbeitsmarktprozesse? Eine institutionentheoretische Skizze. bor Relat Rev 66(1), 225–267 (2013) LASER Discussion Papers, vol. 55. Friedrich- Alexander -Uni- Blien, U., Reinberg, A., Tessaring, M.: Die Ermittlung der Übergänge versität, Erlangen-Nürnberg (2011) zwischen Bildung und Beschäftigung. Mitt Arbeitsmarkt Berufs- Acemoglu, D., Autor, D.: Skills, tasks and technologies: Implications forsch 23(4), 181-204 (1990) for employment and earnings. In: Ashenfelter, O., Card, D. (eds.) Böckermann, P., Ilmakunnas, P.: Job disamenities, job satisfaction, quit Handbook of labor economics, vol. 4, pp. 1043–1171. Elsevier, intentions, and actual separations: putting the pieces together. Ind Amsterdam (2011) Relat (Berkeley) 48(1), 73–96 (2009) Allmendinger, J.: Educational systems and labor market outcomes. Eur Bonin, H., Schneider, M., Qunike, H., Arens, T.: Zukunft von Bildung Sociol Rev 5, 231–249 (1989) und Arbeit. Perspektiven von Arbeitskräftebedarf und -angebot Bacharach, M.: Biproportional matrices and input-output change. bis 2020. IZA Research Report, vol. 9. (2007) Cambridge University Press, Cambridge (1970) Brücker, H., Klinger, S., Möller, J., Walwai, U.: Handbuch Arbeits- Bechmann, S., Dahms, V., Tschersich, N., Frei, M., Leber, U., Schwen- markt 2013. Analysen, Daten, Fakten. IAB Bibliothek, vol. 334. gler, B.: Fachkräfte und unbesetzte Stellen in einer alternden (2013) K Modelling reallocation processes in long-term labour market projections 89 Brunow, S., Garloff, A.: Arbeitsmarkt und demograﬁscher Wandel. IAB-Modellrechnungen zu den Entwicklungen in Berufsfeldern Anpassungsprozesse machen dauerhaften Fachkräftemangel un- und Qualiﬁkationen bis 2025, pp. 103–124. Bundesinstitut für wahrscheinlich. IAB Forum 2/2011. (2011) Berufsbildung, Bielefeld (2010) CEDEFOP: Skills supply and demand in Europe. Medium-term fore- Lapointe, M., Charron, M., Claveau, G., Gendron, M., Gomez, E.G., cast up to 2020. Publications Ofﬁce of the European Union, Lux- Grenier, M., Ignaczak, L., Kim, J.-Y., Lamy, R., Pescarus, C., embourg (2009) Tremblay-Côté, N., Vincent, N., Zou, Y.: Looking-ahead: a 10- CEDEFOP: Future skills and demand in Europe. Forecast 2012. Re- year outlook for the Canadian labour market (2008–2017). Hu- search Paper, vol. 26. Publications Ofﬁce of the European Union, man Resources and Skills Development Canada, Gatineau (2008) Luxembourg (2012) Leontief, W.W.: The structure of American economy, 1919–1939. An Cortes, G.M.: Where have the middle-wage workers gone? A study of empirical application of equilibrium analysis. Oxford University polarization using panel data. J Labor Econ 34(1), 63–105 (2016) Press, Oxford (1951) Cottini, E., Kato, T., Westergaard-Nielsen, N.: Adverse workplace con- Lepic, M., Koucky, J.: Employment forecasting in the Czech Repub- ditions, high-involvement work practices and labor turnover: evi- lic. In: Arendt, L., Ulrichs, M. (eds.) Best practices in forecasting dence from Danish linked employer–employee data. Labour Econ labour demand in Europe. IPISS, Warsaw (2012) 18(6), 872–880 (2011) Lockard, C.B., Wolf, M.: Occupational employment projections to Cotton, J.L., Tuttle, J.M.: Employee turnover: a meta-analysis and 2020. Employment outlook 2010–2020. Mon Labor Rev 135(1), review with implications for research. Acad Manag Rev 11(1), 84–108 (2012) 55–70 (1986) Lucas, R.J., Prescott, E.: Equilibrium search and unemployment. Damelang, A., Schulz, F., Vicari, B.: Institutionelle Eigenschaften von J Econ Theory 7, 188–209 (1974) Berufen und ihr Einﬂuss auf beruﬂiche Mobilität in Deutschland. Maier, T., Afentakis, A.: Forecasting supply and demand in nursing Schmollers Jahrb 135, 307–334 (2015) professions: impacts of occupational ﬂexibility and employment Dupuy, A.: Forecasting expansion demand by occupation and educa- structure in Germany. Hum Resour Health 11(1), 24 (2013). tion in the Netherlands. In: Arendt, L., Ulrichs, M. (eds.) Best doi:10.1186/1478-4491-11-24 practices in forecasting labour demand in Europe, pp. 127–147. Maier, T., Helmrich, R.: Creating the initial vocational qualiﬁcation IPISS, Warsaw (2012) from the German microcensus. Paper presented at the ACSPRI Dustman, C., Glitz, A.: How do industries and ﬁrms respond to changes Conferences, RC33 Eighth International Conference on Social in local labor supply? J Labor Econ 33(3), 711–750 (2015) Science Methodology, Sydney. (2012) Dustmann, C., Ludsteck, J., Schönberg, U.: Revisiting the German Maier, T., Zika, G., Mönnig, A., Wolter, M.I., Kalinowski, M., Hänisch, wage structure. Q J Econ 124(2), 843–881 (2009) C., Helmrich, R., Schandock, M., Neuber-Pohl, C., Bott, P., Hum- Ehing, D., Moog, S.: Erwerbspersonen- und Arbeitsvolumenprojektio- mel, M.: Wages and occupational ﬂexibilities as determinants of nen bis ins Jahr 2060. J Labour Mark Res 46(2), 167–182 (2013) the interactive QuBe labour market model. Discussion Papers Fitzenberger, B., Lickleder, S., Zwiener, H.: Mobility across ﬁrms and No. 149. Federal Institute for Vocational Education and Training, occupations among graduates from apprenticeship. IZA Discus- Bonn (2014) sion Paper, vol. 9006. (2015) Maier, T., Mönnig, A., Zika, G.: Labour demand in Germany by in- Fitzenberger, B., Spitz-Oener, A.: Die Anatomie des Berufswechsels: dustrial sector, occupational ﬁeld and qualiﬁcation until 2025 – Eine empirische Bestandsaufnahme auf Basis der BIBB/IAB- model calculations using the IAB/INFORGE model. Econ Syst Daten 1998/1999. In: Franz, W., Ramser, H.J., Stadler, M. (eds.) Res 27(1), 19–42 (2015) Bildung Wirtschaftswissenschaftliches Seminar in Ottobeuren, Mayer, K.U., Carroll, G.R.: Jobs and classes: structural constraints on vol. 33, pp. 29–54. Mohr Siebeck, Tübingen (2004) career mobility. Eur Sociol Rev 3(1), 14–38 (1987) Gajdos, A., Zmurkow-Poteralska, E.: Employment forecasts by oc- McLaughlin, K.J.: A theory of quits and layoffs with efﬁcient turnover. cupational groups in Poland. Polit Spoleczna Themat 2014(2), J Polit Econ 99(1), 1–29 (1991) 14–20 (2014) Meyer, B., Lutz, C., Schnur, P., Zika, G.: National economic policy Gathmann, C., Schönberg, U.: How general is human capital? A task- simulations with global interdependencies. A sensitivity analysis based approach. J Labor Econ 28, 1–49 (2010) for Germany. Econ Syst Res 19(1), 37–55 (2007) Geel, R., Backes-Gellner, U.: Occupational mobility within and be- Montgomery, J.D.: Equilibrium wage dispersion and interindustry tween skill clusters: an empirical anaysis based on the skill- wage differentials. Q J Econ 106(1), 163–179 (1991) weights approach. Empir Res Vocat Educ Train 3, 21–38 (2011) Moscarini, G., Thomsson, K.: Occupational and job mobility in the Gibbons, R., Katz, L.F.: Layoffs and lemons. J Labor Econ 9(4), US. Scand J Econ 109(4), 807–836 (2007) 351–380 (1991) Nisic, N., Trübswetter, P.: Berufswechsler in Deutschland und Großbri- Goos, M., Manning, A., Salomons, A.: Job polarization in Europe. Am tannien. IAB-Kurzbericht 1/2012. (2012) Econ Rev 99(2), 58–63 (2009) OECD: Managing decentralisation – a new role for labour market pol- Groes, F., Kircher, P., Manovskii, I.: The U-shape of occupational mo- icy. OECD Publications Service, Paris (2003) bility. Rev Econ Stud 82(2), 659–692 (2015) Papps, K.L.: Occupational and skill forecasting: a survey of overseas Grossman, G.M., Rossi-Hansberg, E.: Trading tasks: a simple theory approaches with applications for New Zealand. Occasional Paper of offshoring. Am Econ Rev 98(5), 1978–1997 (2008) 2001/1. Labour Market Policy Group, Wellington (2001) Helmrich, R., Zika, G. (eds.): Beruf und Qualiﬁkation in der Zukunft. Pissarides, C.A.: Equilibrium unemployment theory. MIT Press, Cam- BIBB-IAB-Modellrechnungen zu den Entwicklungen in Berufs- bridge (2000) feldern und Qualiﬁkationen bis 2025. Bundesinstitut für Berufs- Pollmann-Schult, M.: Ausmaß und Struktur von arbeitnehmerinduzierter bildung, Bielefeld (2010) Abstiegsmobilität. Kolner Z Soz Sozpsychol 4, 573–591 (2006) Helmrich, R., Maier, T.: Employment forecasting in Germany – an oc- Reichelt, M., Abraham, M.: Occupational and regional mobility as sub- cupational ﬂexibility matrix approach. In: Arendt, L., Ulrichs, M. stitutes. A new approach to understanding job changes and wage (eds.) Best practices in forecasting labour demand in Europe In- inequality. IAB-Discussion Paper 14/2015. (2015) stytut Pracy I Spraw Socialnych, Report II. pp. 103–126. (2012) Schnur, P., Zika, G.: Das IAB/INFORGE-Modell. Ein sektorales Hirschmann, A.O.: The paternity of an index. Am Econ Rev 54(5), 761 makroökonomisches Projektions- und Simulationsmodell zur (1964) Vorausschätzung des längerfristigen Arbeitskräftebedarfs. IAB- Kalinowski, M., Quinke, H.: Projektion des Arbeitskräfteangebots bis Bibliothek 318. Institut für Arbeitsmarkt und Berufsforschung, 2025 nach Qualiﬁkationsstufen und Berufsfeldern. In: Helmrich, Nürnberg (2009) R., Zika, G. (eds.) Beruf und Qualiﬁkation in der Zukunft. BIBB- K 90 T. Maier et al. Caroline Neuber-Pohl is researcher at the German Federal Insti- Shavit, Y., Müller, W.: Vocational secondary education, tracking, and tute for Vocational Education and Training (BIBB), Section “Qualiﬁca- social stratiﬁcation. In: Hallinan, M.T. (ed.) Handbook of sociol- tion, Occupational Integration, Employment”, since 2015. She holds an ogy of education, pp. 437–452. Springer, New York (2000) MSc in Economics, Rheinische Friedrich-Wilhelms-University Bonn, Shaw, J.D., Delery, J.E., Jenkins, G.D., Gupta, N.: An organiza- 2015 and is doctoral student at the Friedrich-Alexander-University, Er- tion-level analysis of voluntary and involuntary turnover. Acad langen-Nürnberg. Her research areas are: Labour market adjustments, Manag J 41(5), 511–525 (1998) Spitz-Oener, A.: Technical change, job tasks, and rising educational de- occupational integration, vocational education. mands. Looking outside the wage structure. J Labor Econ 24(2), Anke Mönnig studied Economics at the Free University of Berlin 235–270 (2006) and holds a Master of Arts Degree in International Economics from Tiainen, P.: Employment forecasting in Finland. In: Arendt, L., Ulrichs, the Berlin School of Economics and Law. In 2006 she joined the In- M. (eds.) Best practices in forecasting labour demand in Europe, stitute for Economic Structures Research (GWS) in Osnabrück. Since pp. 51–62. IPISS, Warsaw (2012) 2016, she is deputy head of the GWS division “Economic and Social Tiemann, M., Schade, H.-J., Helmrich, R., Hall, A., Braun, U., Bott, Affairs”. Her thematic focus concentrates on industrial analysis, labor P.: Berufsfeld-Deﬁnitionen des BIBB auf Basis der KldB1992. markets and structural change with an emphasis on external trade. Wissenschaftliche Diskussionspapiere No. 105. Bundesinstitut She is responsible for the macroeconometric input-output model IN- für Berufsbildung, Bonn (2008) UK Commission for Employment and Skills: Working Futures 2010–2020. FORGE (INterindustry FORcasting Germany) and the world trade Executive Summary, vol. 41. UK Commission for Employment model TINFORGE. and Skills, South Yorkshire (2011) Dr. Gerd Zika studied Business Management at the Friedrich-Alexan- Vogler-Ludwig, K., Düll, N.: Arbeitsmarkt 2030. Eine strategische der University Erlangen/Nuremberg (M. Sc. 1991). Then he worked as Vorausschau auf Demograﬁe, Beschäftigung und Bildung in an assistant at the Chair of Statistics and Econometrics at the Friedrich- Deutschland. Bertelsmann, Bielefeld (2013) Alexander-University Erlangen/Nuremberg (Dr. rer. pol. 1994). In Weber, M.: Wirtschaft und Gesellschaft. Mohr Siebeck, Tübingen 1995 he joined the Institute for Employment Research (IAB), where (1972) Whelpton, P.K.: An empirical method of calculating future population. he is responsible for the BIBB-IAB-Qualiﬁcation and Occupational J Am Stat Assoc 31(195), 457–473 (1936) Field Projections (QuBe-Projekt.de) since 2007. His thematic focus Wilson, R.: Forecasting skill requirements at national and company concentrates on the analysis of both short- and long-term developments level. In: Descy, P., Tessaring, M. (eds.) Training in Europe. Sec- in the labor market, with a main focus on the labor demand side. ond report on vocational training research in Europe 2000. Back- Michael Kalinowski studied Economics at the University of Regens- ground report, pp. 561–609. Ofﬁce for Ofﬁcial Publications of the burg (Dipl. Vw./MSc in Economics). In 2007 he joined as researcher European Communities, Luxembourg (2001) the Fraunhofer Institute for Applied Information Technology (Fraun- Zika, G., Helmrich, R., Kalinowski, M., Wolter, M.I., Hummel, M., hofer FIT) in Sankt Augustin, Germany, research group MikMod. Maier, T., Hänisch, C., Drosdowski, T.: In der Arbeitszeit steckt Since 2016 he is researcher at the Federal Institute for Vocational noch eine Menge Potenzial. Qualiﬁkations- und Berufsfeldpro- Education and Training (BIBB) in Bonn, Germany, in the Section jektionen bis 2030. IAB-Kurzbericht 18/2012. (2012) “Qualiﬁcations, Occupational Integration and Employment”. His re- search interests focus on economics of education and labour market Tobias Maier studied Politics and Management at the University economics including projection of the future qualiﬁcation and occupa- of Constance (Dipl. Verw.-Wiss./Master of Arts). In 2009, he joined tion structure of the population and labour work force. the Federal Institute for Vocational Education and Training (BIBB) in Bonn, Germany, as Researcher in the Section “Qualiﬁcations, Occupa- tional Integration and Employment”. He is responsible for the BIBB- IAB-Qualiﬁcation and Occupational Field Projections (QuBe-Pro- jekt.de) and the econometric estimation and simulation of the yearly supply and demand of vocational training (PROSIMA). His further research areas are: educational development, vocational educational, access to employment and labour mobility.

Journal

Journal for Labour Market Research – Springer Journals

Published: Apr 5, 2017

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Modelling reallocation processes in long-term labour market projections

Modelling reallocation processes in long-term labour market projections

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Modelling reallocation processes in long-term labour market projections

Modelling reallocation processes in long-term labour market projections

References (62)

Abstract

Journal

Recommended Articles

There are no references for this article.

Our policy towards the use of cookies