Show Summary Details

Page of

Printed from Oxford Research Encyclopedias, Economics and Finance. Under the terms of the licence agreement, an individual user may print out a single article for personal use (for details see Privacy Policy and Legal Notice).

date: 26 February 2024

Business Cycles and Apprenticeshipsfree

Business Cycles and Apprenticeshipsfree

  • Samuel MuehlemannSamuel MuehlemannMunich School of Management, Ludwig-Maximilians-Universität München
  •  and Stefan WolterStefan WolterFaculty of Business, Economics and Social Sciences, University of Bern


The economic reasons why firms engage in apprenticeship training are twofold. First, apprenticeship training is a potentially cost-effective strategy for filling a firm’s future vacancies, particularly if skilled labor on the external labor market is scarce. Second, apprentices can be cost-effective substitutes for other types of labor in the current production process. As current and expected business and labor market conditions determine a firm’s expected work volume and thus its future demand for skilled labor, they are potentially important drivers of a firm’s training decisions.

Empirical studies have found that the business cycle affects apprenticeship markets. However, while the economic magnitude of these effects is moderate on average, there is substantial heterogeneity across countries, even among those that at first sight seem very similar in terms of their apprenticeship systems. Moreover, identification of business cycle effects is a difficult task. First, statistics on apprenticeship markets are often less developed than labor market statistics, making empirical analyses of demand and supply impossible in many cases. In particular, data about unfilled apprenticeship vacancies and unsuccessful applicants are paramount for assessing potential market failures and analyzing the extent to which business cycle fluctuations may amplify imbalances in apprenticeship markets. Second, the intensity of business cycle effects on apprenticeship markets is not completely exogenous, as governments typically undertake a variety of measures, which differ across countries and may change over time, to reduce the adverse effects of economic downturns on apprenticeship markets. During the economic crisis related to the COVID-19 global pandemic, many countries took unprecedented actions to support their economies in general and reacted swiftly to introduce measures such as the provision of financial subsidies for training firms or the establishment of apprenticeship task forces. As statistics on apprenticeship markets improve over time, such heterogeneity in policy measures should be exploited to improve our understanding of the business cycle and its relationship with apprenticeships.


  • Health, Education, and Welfare Economics
  • Labor and Demographic Economics


Company-based apprenticeship training has been touted by international organizations and many governments for more than two decades (e.g., Organisation for Economic Co-operation and Development [OECD], 1999, 2010) as a means of combating often persistently high rates of youth unemployment—first, because young people working in companies acquire skills in demand in the labor market in the short term (which prevents mismatch), and second, because those young people simultaneously gain practical experience during their training, which facilitates their transition to the labor market without needing to spend extra time in poorly paid internships. Particularly in economically difficult times, when job offers are scarce, people with an apprenticeship qualification may have a significant advantage in finding a job over young people with more general and theory-based types of education. These initial advantages when entering the labor market, which are well documented in the literature (e.g., Hanushek et al., 2017), are offset by the challenge of finding enough companies to provide training places, particularly during times of economic downturns. As the number of apprenticeships depends on both the number of individuals and the number of firms willing to train them, many more individuals may be looking for training places when firms are offering fewer training places during a recession and vice versa during an economic expansion.

We limit our discussion to apprenticeships in which a firm not only provides part of the training but also makes the recruitment decision. In many apprenticeship systems, the decisions regarding the number of training opportunities in particular training occupations are made by schools, which then look for work placements with firms for the later part of the training. It is quite conceivable that such training models would also be affected by economic cycles, as vocational schools must ensure sufficient work placements for their apprentices. However, the enrollment policy of vocational schools likely depends less on business cycle fluctuations than on a firm’s decision to hire apprentices. Work placements for individuals enrolled in vocational schools are typically of a rather short duration (a few weeks or months) and paid rather poorly or not at all so that offering such placements does not constitute a financial investment for the host companies. In addition, the revenue of vocational schools directly depends on their number of students, which creates incentives to keep enrollment fluctuations at a minimum.

Although the school-based education system is not independent of the business cycle, as fiscal austerity measures could force states to reduce spending on education or individuals may no longer be able to afford tuition, school-based vocational education should be less sensitive to the business cycle than vocational training with a strong workplace component (Méndez & Sepúlveda, 2012). Nonetheless, both the theoretical and empirical literature on how the business cycle affects vocational education and training has remained surprisingly small, and empirical studies are limited in terms of the time periods and geographical regions examined. Thus, it is still not conclusively clear how strongly business cycle fluctuations affect apprenticeship markets.

This article not only provides an overview of the theoretical foundations and empirical work on this topic but also addresses the unanswered questions that often arise due to insufficient or missing empirical data that would be necessary to identify the association between the business cycle and apprenticeship market outcomes. Following a theoretical introduction, data and measurement issues are discussed. As the early literature was reviewed in Brunello (2009), our review of empirical studies focuses mainly on the two periods since the Great Recession and includes a discussion of the insights gained in the context of the COVID-19 global pandemic, followed by conclusions that include some policy implications and avenues for future research.

Theoretical Observations

A market for apprenticeship training can be characterized in many ways as a labor market and is thus subject to supply and demand. Firms post vacancies for apprenticeships, to which interested individuals apply. Changes in the number of newly concluded apprenticeship contracts each year are therefore the result of changes in demand and/or supply caused by the business cycle, in addition to several other factors (such as demographic or technological change).

The Impact of Business Cycles on the Demand for Apprentices

To understand the influence of the business cycle on the demand for apprentices, it is helpful to address the question of why firms train apprentices in the first place. Human capital theory predicts that firms are not willing to invest in general human capital if labor markets are competitive, although individuals and firms would share the training costs and benefits from firm-specific investments (for a more detailed discussion, see Muehlemann & Wolter, 2020; Wolter & Ryan, 2011). While these theories have been refined and expanded over time, two basic motives for apprenticeship training have emerged from the perspective of firms. Firms may see a short-term benefit in apprenticeship training because they can substitute both unskilled and skilled workers with less costly apprentices. The main reason for apprenticeship training is thus to make production more cost-effective, which is why this motivation is also called a production-oriented training strategy. For such a strategy to be profitable, the value of an apprentice’s productive contribution must cover a firm’s training expenditures, which means that apprentice wages must be sufficiently low. However, firms may also expect returns to training in the medium or long run, in which case the benefit that apprentices create during training becomes less important. Depending on the magnitude of the expected post-training benefits, firms may be willing to make an initial net investment in apprenticeship training. Such a training strategy is therefore also called investment-oriented training motivation. A potentially important type of post-training benefit is savings on hiring costs from not having to hire skilled workers from the external labor market. In this case, training apprentices and subsequently retaining them as skilled workers can be more cost-effective than hiring skilled workers from the labor market. This may be the case if the net training investment in apprentices is lower than the costs of recruiting and onboarding suitable skilled workers from the external labor market (Blatter et al., 2012, 2016).

Business cycle fluctuations may affect a firm’s training behavior differently depending on the firm’s training motivation. In the case of production-oriented firms, it is conceivable that an economic downturn would increase the incentives for a firm to substitute skilled workers with cheaper apprentices to further reduce production costs (Stevens, 1999). However, to the extent that training regulations in a particular country (or occupation) restrict a firm’s ability to use apprentices in the production process, the potential benefit from using additional apprentices as a cost-effective production factor may be quickly offset by training costs.1 Furthermore, a reduction in skilled workers for the purpose of substitution with cheaper apprentices may be difficult in countries with strongly regulated labor markets (i.e., a high degree of employment protection legislation; see OECD, 2020). Finally, production-oriented training firms are able to generate a net benefit by the end of the training period (see, e.g., Muehlemann & Wolter, 2014). The value of an apprentice’s productive tasks, however, may decrease during a recession if firms do not have enough work volume to use apprentices in the production process so that firms may no longer be able to offer training profitably. However, as firms are aware of business cycle fluctuations in general, a production-oriented training strategy provides a buffer against short-term drops in the volume of work for apprentices. In other words, to the extent that training firms expect an economic downturn to be limited in time and not too severe, they are able to absorb its negative effects, and such a downturn is therefore likely to have only a marginal impact on their willingness to hire new apprentices. In summary, although companies with a production-oriented training strategy are likely to hire fewer new apprentices in response to a decrease in their volume of work and vice versa, the identification of the magnitude of a firm’s reaction is left to empirical analyses.

At first glance, one might assume that the training decisions of firms with investment-oriented training motives would be less sensitive to the business cycle than those of firms with a production-oriented training motive. However, this need not be the case for at least two reasons. First, according to the investment motive, training firms are willing to accept certain net training costs because of expected post-training benefits from retaining apprentices as skilled workers after training. However, if an economic downturn is too strong and the net training costs become too high to be realistically compensated later, then a reduction in training offers could also occur here. Second, as in the case of production-oriented firms, much depends on the expectations of the firms. If firms expect a prolonged crisis that will also have a delayed effect on the labor market, then firms may assume that recruitment costs will also decrease for a certain period of time (see, e.g., Muehlemann & Strupler Leiser, 2018) and that the decrease in post-training benefits may no longer compensate for the increase in net training costs. However, in the case of a rather short economic downturn, being able to retain apprentices during the next boom period may in fact lead to even higher post-training benefits to the extent that labor market tightness increases accordingly. Conversely, if a firm must reduce its workforce due to an economic downturn, then it will not be able to retain as many apprentices as originally planned and thus be unable to recoup part of its training investment of apprentices that it already hired before the start of a recession.

Finally, bankruptcies triggered by an economic downturn may cause training firms to disappear from the market. Because bankrupt firms usually cannot be quickly replaced by firms that have not yet been active in training and because other training firms are unlikely to increase their number of apprenticeships during a recession, business cycle–related bankruptcies are another channel that could decrease the demand for apprentices independently of a firm’s training strategy.

In summary, an economic downturn likely reduces a firm’s demand for apprentices regardless of its training motive. However, the effect of business cycle fluctuations on the apprenticeship market depends on a firm’s expectations about the magnitude and duration of the business cycle itself and on the nature of the regulations in a country’s labor and apprenticeship markets. Finally, business cycle effects may be asymmetric (i.e., adverse effects during economic downturns may differ in magnitude from the positive effects during boom periods).

The Impact of Business Cycles on the Supply of Apprentices

The number of new apprenticeships may depend as much on the willingness of individuals to undergo training as on the willingness of companies to offer apprenticeships. Many empirical studies find a strong association between demographic changes and outcomes in the apprenticeship market (e.g., Lüthi & Wolter, 2020a; Muehlemann et al., 2020). However, the extent to which the supply of apprentices is affected by the business cycle is less clear. These effects are likely to vary depending on the function of apprenticeship training in an education system. In German-speaking countries, apprenticeship training is predominantly a type of upper secondary education for those graduating from the compulsory school system and thus a viable alternative to attending college or university. Conversely, in other countries, such as the United States, apprenticeship training is aimed more at young adults in their late twenties who, often after unemployment, try to reorient themselves and regain a foothold in the labor market. Thus, depending on the target participants in an apprenticeship system, the educational options for potential apprentices are different in the event of an economic boom or a recession. Some can easily opt for a different type of training; for example, enrolling at a university, opting for a job in the labor market, or postponing their training by settling for a temporary interim solution. It is therefore conceivable that changes in the demand for apprentices affect the target participants differently depending on their alternative educational and employment opportunities. Furthermore, the responses of supply to economic fluctuations are likely asymmetrical. In a recession, individuals may be forced to enroll in school-based educational tracks because firms reduce their demand for apprentices. Conversely, during an economic boom period, the opportunity costs of training increase when labor markets are tight (and wages increase), particularly for individuals who are already employed as semiskilled workers.

In summary, although an economic downturn mainly affects an apprenticeship market by decreasing a firm’s demand for apprentices, it can be assumed that business cycle fluctuations also affect the supply of apprentices through their impact on individuals’ opportunity costs of training. However, such effects are likely less relevant in systems where apprenticeship training is an important educational track at the upper secondary level.

Other Actors in the Apprenticeship Market

Just like labor markets, apprenticeship markets may be subject to interventions by the state, particularly during economic downturns.2 On the one hand, such interventions can cushion adverse effects by offering alternative training opportunities to unsuccessful applicants, either temporarily as interim solutions or permanently through the creation of school-based training.3 Moreover, the state can also intervene directly by trying to stimulate demand through support measures (e.g., through subsidizing apprenticeships or lowering the costs of training by allowing companies to claim tax deductions). In addition, there are also several possible nonmonetary measures, such as marketing campaigns and appeals to the social responsibility of companies (Schweri et al., 2019). Finally, it is also important to consider stimulus measures that benefit all companies that support the economy in general and thus indirectly the willingness of companies to train. During the COVID-19 global pandemic, many countries put a temporary ban on debt collection to prevent company bankruptcies. A positive side effect of such measures was that companies that may have gone bankrupt in the absence of such measures could continue training their previously hired apprentices and were possibly even willing and able to take on new apprentices in 2020. Thus, it is important to note that the reaction of the apprenticeship market to cyclical fluctuations is at least in part a result of such external interventions, although evaluating their effectiveness is often not feasible (for an exception, see Federal Ministry of Labour and Social Affairs, 2013). Moreover, labor market institutions affect a firm’s hiring and firing decisions. To the extent that employment protection legislation (cf. OECD, 2020) makes layoffs more costly, firms may be reluctant to lay off employees immediately following a macroeconomic shock, particularly when the expected duration is rather short. Moreover, training and subsequently retaining apprentices may be more beneficial for firms in countries with strong employment protection legislation, as apprenticeship training can be used as a screening device to avoid costly mismatches when hiring from the external labor market (Muehlemann et al., 2010). Understanding the reactions of the apprenticeship market to cyclical fluctuations therefore also requires consideration of the actions of the state, labor market and training institutions, and other actors.

Data and Measurement Issues

Modern research in labor economics is based on search and matching models that explicitly model vacancies and job seekers to better understand the empirical fluctuations in labor market outcomes (Pissarides, 2000). Data on employment and wages are usually available for the entire workforce, but information on job vacancies and job seekers is incomplete. First, not all vacancies are advertised on job portals or company websites or registered with employment agencies. Moreover, some firms may no longer advertise vacancies if they expect that they will not be successful in finding suitable applicants, while others post vacancies but have no intention of hiring anybody unless an exceptional candidate applies or simply do not bother to remove the ad once the vacancy is filled. Second, not all individuals who are actively looking for work are registered as unemployed because some unsuccessful job seekers receive unemployment benefits for a limited time and others look for new jobs while being employed (job-to-job searches). Similar issues arise in the context of apprenticeship training.

Measuring the Demand for and Supply of Apprentices

A firm’s demand for apprentices is defined as the sum of its apprenticeship contracts and unfilled training positions. Data on the number of training contracts are typically available for the universe of training positions at the state level. However, information about the number of unfilled apprenticeship positions is often incomplete.4

The supply of apprentices is the sum of the number of successful applicants who sign a training contract and the number of unsuccessful applicants. The number of unsuccessful applicants is even more difficult to measure empirically than the number of unfilled apprenticeship vacancies. On the one hand, many potential apprentices apply for training positions immediately after graduating from compulsory schooling (particularly in countries such as Germany and Switzerland); thus, the number of school leavers (by education level) is an obvious proxy for the potential supply of apprentices to a firm. However, in many apprenticeship countries, the number of older individuals who apply for apprenticeships after having participated in the labor market for some time has increased in recent years. Such individuals include semiskilled workers with no postcompulsory schooling qualifications or individuals who have changed (or had to change) their occupation and thus desire to acquire another formal vocational qualification. Moreover, initially unsuccessful applicants for training positions may enroll in preparatory courses in a transition system that is supposed to fill skills gaps to improve an individual’s chances on the apprenticeship market in the subsequent year, may opt for entirely school-based education programs, may enter the workforce as semiskilled employees, may become unemployed, or may decide not to pursue any further education and withdraw from the labor market.

One option for measuring the supply of apprentices is to include all individuals who were interested in a training position at some point in a particular year.5 However, using the number of interested individuals would overestimate actual supply because they might have ended up changing their minds and preferred to opt for a school-based education (e.g., enrolling in a university or community college). A second option is to focus on individuals who are still registered as actively looking for an apprenticeship position with an employment agency by the time the new school year starts. This approach, however, would underestimate the true number of unsuccessful applicants, as some might have opted for a second-best option and enrolled in the transitory system.6 A third option is to rely on firm-level panel data and gather information on the number of applicants for advertised training positions. To the extent that representative firm-level panel data sets are available, this option would likely generate the most accurate picture regarding the association between supply and apprenticeship market outcomes by accounting for heterogeneity at the firm level. However, while firm-level panel data sets are available in some countries, the number of observations is typically too small to carry out a meaningful analysis within (smaller) occupations while accounting for regional factors.7 Moreover, some of the outcome variables of interest often do not change over time (i.e., many small firms train only one apprentice at a time), making identification difficult.

Measurement Issues Regarding Business Cycles

A final measurement issue is related to the business cycle itself. The predominant measure of economic activity is gross domestic product (GDP), as it captures much of the economic activity in a country and the relevant changes therein over time. Fluctuations in GDP are referred to as the business cycle, with boom periods occurring when GDP is higher than in previous periods and recessions when GDP growth is negative (typically for two consecutive quarters). Other indicators reflect tightness in the labor market, which is associated with (and typically lags behind) the business cycle, as firms post more vacancies and unemployment is lower during boom periods. In labor market theory (Pissarides, 2000), the relevant indicator for labor market tightness is the ratio of these variables (i.e., the vacancy-unemployment ratio), although many empirical studies include only the unemployment rate as a proxy for the business cycle because representative vacancy data are often not available.8

Firms’ decisions to offer training places for apprentices are based on their expectations about the business cycle in the (not-so-distant) future rather than on the current situation for at least two reasons: first, because apprenticeships typically require an investment period of several years, and second, because training vacancies are usually put on the market up to a year or more before training actually starts.9

Empirically, there are several possibilities for measuring business expectations. The most used option is to directly measure firms’ business expectations by means of survey data. Different indicators are available in different countries, and they typically correlate strongly with subsequent GDP measures (e.g., Sauer & Wohlrabe, 2020, report for Germany that the Ifo Business Climate Indicator (BCI) leads GDP by approximately one to two quarters).10 Other indicators have been developed to specifically capture the situation in the labor market regarding employment (vacancies) and unemployment.

To the extent that a firm’s expectations are true on average, actual realizations of GDP in a particular period may be used in place of unobserved expectations for GDP in ex post analyses but of course not for ex ante forecasts.11 Except for Dietrich and Gerner (2007) and Muehlemann et al. (2020), all empirical studies use current and/or lagged GDP and/or unemployment as business cycle measures.12

Another complication is that firms typically hire apprentices only once a year. Hiring often occurs in the first two quarters of the year because a large majority of apprentices in countries such as Austria, Germany, Switzerland, and the United Kingdom start their training in August or September. Thus, using lagged measures of GDP also accounts to some degree for the uncertain timing of when apprenticeship contracts are signed throughout the year. Some firms have already signed apprenticeship contracts up to a year prior to the start of an apprenticeship, which implies that the business cycle a year prior to the start of training—not GDP in the starting year of the apprenticeship—potentially impacted the number of advertised apprenticeship vacancies. In reality, the situation is certainly quite heterogeneous. At the time of a macroeconomic shock, some firms may have already signed training contracts while others may not have hired any apprentices or not yet successfully filled all available training positions.

Finally, depending on a country’s training regulations, firms that signed a contract prior to an economic shock may still dissolve that contract either prior to the start of training or during the probationary period (which usually lasts up to several weeks).

In summary, data that capture business expectations can provide timely information that is relevant to a firm’s training decisions because such data are often available on a monthly basis and made public shortly after the survey takes place. Therefore, such data can be used to forecast developments in the training market in the (near) future. In the absence of data on expectations and to the extent that a firm’s expectations are rational, researchers can of course also make use of actual outcomes of GDP and labor market variables such as employment, vacancies, or unemployment (with the appropriate time structure).13

A further important distinction regarding business cycle measures is whether they should be ascertained at the national, industry, and/or regional level. Clearly, some macroeconomic shocks affect the entire economy simultaneously. Many macroeconomic shocks, however, affect some industries more strongly than others. Additionally, the shock caused by the COVID-19 pandemic heavily affected some industries (e.g., tourism, restaurants, and hotels) due to repeated lockdowns and travel bans, while companies in other sectors remained open for business. Thus, firms’ expectations about their future demand for labor may differ quite substantially across industries. As a result, to the extent that macroeconomic shocks are idiosyncratic across industries, industry-level business cycle measures may contain very important information for explaining the within-industry development of apprenticeship markets.

Business Cycles vs. Structural Changes and Sector-Specific Crises

Neither firms nor politicians nor experts know exactly which part of an economic downturn is only cyclical in nature and thus limited in duration (temporary) and which part is structural and affects markets permanently, and structural changes in the economy are often triggered by cyclical fluctuations. The distinction, however, is crucial because firms are unlikely to alter their demand for apprentices if they expect only a short cyclical downturn, due to the long-term nature of investments in apprenticeship training. If, on the other hand, a downturn is expected to have a structural component, then there will also be longer-term shifts in the demand for certain occupations and skills. These shifts make larger investments in skills that are becoming obsolete no longer worthwhile. As a result, both supply and demand in the apprenticeship market are expected to react accordingly (i.e., reactions to an economic downturn that affects the long-term structure of the economy would be more severe). In the case of the economic crisis triggered by the COVID-19 pandemic, it was originally expected that there would be a short cyclical movement that would not actually have a major impact on the apprenticeship market. However, it is unclear whether this crisis also contains certain structural components. For example, experiences with video conferencing and working from home may have a lasting impact on business travel, which in turn would have an adverse impact on jobs in transportation, hotels, and restaurants and accordingly on the demand for and supply of apprentices in these sectors in the medium term.

Finally, it must be considered that the apprenticeship system in many countries covers only certain sectors and occupations and that economic and structural changes can affect sectors very differently. An economic downturn can be heavily concentrated in only a few sectors (e.g., the real estate and construction sectors were the most heavily affected in many countries during the Great Recession). Thus, if an apprenticeship system is also concentrated in precisely these same sectors, a downturn would have a tremendously strong negative effect compared to a downturn of the same size that affects other sectors offering few or no apprenticeships. It is therefore important to consider such heterogeneity regarding business cycle effects and the prevalence of apprenticeships. In addition, countries with apprenticeship systems that are spread across many sectors are generally better protected against economic shocks because sectors that are less affected can compensate for the reduction in demand for apprentices in the more affected sectors. Conversely, if such compensatory opportunities do not exist, then a major cyclical shock can either have no effect at all or can practically destroy an apprenticeship system overnight.

Identification Strategies

Longitudinal data are typically required to identify business cycle effects due to the dynamic nature of macroeconomic shocks, although it may be possible to exploit regional variation in economic activity at a particular point in time. As the level of apprenticeship training differs widely across certain occupational groups, industries, and regions, empirical studies typically estimate panel data models that allow time-invariant unobservable factors to be accounted for.14 Empirical studies usually focus on the number of apprenticeship contracts. However, we might expect the business cycle to have different effects on the demand for and supply of apprentices. Thus, ideally, business cycle effects on the demand for and the supply of apprentices should be estimated separately. However, due to a lack of appropriate data on unfilled vacancies and unsuccessful applicants, such estimations are often not feasible. Moreover, factors such as technological change, demographic change, regional educational policy, changes in the minimum wage for apprentices, training subsidies, changes in migration policies or shifting occupational preferences among young people may simultaneously affect the demand for and supply of apprentices over time.15 To the extent that such unobserved factors are time variant, the resulting coefficients on business cycle measures would be biased. However, including trend variables at the regional level may at least partly account for long-term shifts in educational preferences, which may be heterogeneous at the regional level because education systems, including those of Germany and Switzerland, are often not centralized. Furthermore, industry-level variables may to some degree capture gradual (not disruptive) technological change, which in turn affects how firms produce goods and services and thus impacts a firm’s future demand for skilled workers and apprentices (Acemoglu & Restrepo, 2019).

Empirical Studies

Brunello (2009) provided an extensive review of the early literature on apprenticeship training and its association with the business cycle before the Great Recession (2007–2009). We thus refer the reader to that review for details and focus mainly on the more recent literature that analyzes the effects of the Great Recession and the COVID-19 global pandemic. A detailed overview of all relevant empirical studies is provided in Table 1. Institutional features of the corresponding countries where the studies were conducted are summarized in Table 2.

Business Cycle Effects Prior to the Great Recession

For Germany, Dietrich and Gerner (2007) analyzed the effects of business cycle expectations during the period 1993–2003 based on the establishment panel from the Institute for Employment Research (IEB). They found that a one-percentage-point increase in business cycle expectations increased the number of new apprenticeship positions in a firm by 0.4% (based on a panel fixed-effects model) or 0.35% (based on a panel autoregressive distributional lag [ADL] model).16 For the Swiss apprenticeship market, Schweri and Müller (2008) used establishment-level data for the years 1995, 1998, 2001, and 2005. They found procyclical effects of the business cycle, as measured by within-sector GDP growth, on both firm training participation and training intensity (i.e., the number of apprentices in relation to all employees). Muehlemann et al. (2009) used panel data at the canton level from 1988 to 2004 and regressed the number of apprenticeships on cantonal real income and unemployment. They found statistically significant but economically small business cycle effects on the number of apprenticeships. For the years 1990–1996 in Norway, Askilden and Nilsen (2005) and subsequently Brunello (2009), for the period 2001–2007, found that the number of apprenticeship contracts was negatively related to the current and future unemployment rate.

Table 1. Empirical Studies on Apprenticeship Market Outcomes and the Business Cycle



Time period


Dependent variable

Business cycle measure



Baldi et al. (2014)


State-level panel data

Number of new apprenticeship contracts

State-level unemployment rate and income growth

Procyclical effects of the business cycle

Bellmann et al. (2014)


IAB establishment panel

Training decision, training intensity

Establishment-level assessment of how the crisis impacted the establishment

Procyclical effects of the business cycle

Dietrich and Gerner (2007)


IAB establishment panel

Number of new apprenticeship contracts

Establishment-level business expectation

Procyclical effects of the business cycle

Maier (2020)


Yearly economic indicators, labor market outcomes, and school graduates

Demand, supply, number of new apprenticeship contracts

GDP, labor market indicators

Procyclical effects of the business cycle

Muehlemann et al. (2020)


Occupation-state-level panel data

Demand, supply, number of new apprenticeship contracts

ifo Business Climate Index, ifo Employment Barometer

Procyclical effects of the business cycle


Askilden and Nilsen (2005)


Firm-level data in manufacturing and construction industry

Cost shares of apprentices

Unemployment rate

Procyclical effects of the business cycle

Brunello (2009)


National statistics

Ratio of apprentices to pupils in VET schools

Unemployment rate

Procyclical effects of the business cycle


Muehlemann et al. (2009)


State-level panel data

Number of new apprenticeship contracts

State-level unemployment rate and income growth

Procyclical effects of the business cycle

Lüthi and Wolter (2020a)


State-level panel data

Number of new apprenticeship contracts

State-level unemployment rate and income growth

Procyclical effects of the business cycle

Schweri and Müller (2008)

1995, 1998, 2001, 2005

Establishment-level data

Number of new apprenticeship contracts

GDP growth (average of past three years)

Procyclical effects of the business cycle

United Kingdom

Felstead et al. (2013)


Quarterly labor force survey

Share of apprentices

No explicit business cycle measure, recession identified by time period (years)

Procyclical effects of the business cycle

Ventura (2020)


Individualized learner records

Number of new apprenticeship contracts

No explicit business cycle measure, recession identified by time period (years)

Procyclical effects of the business cycle for young apprentices, anticyclical effects for individuals aged 25+

Notes: IAB, Institute for Employment Research Nuremberg; ifo, ifo Institute Munich; VET, vocational education and training.

Effects of the Great Recession on Apprenticeships

A few studies have estimated the effect of the Great Recession on apprenticeships. For Germany, Bellmann et al. (2014) used data from the IEB establishment panel and classified firms based on their subjective assessment of whether each firm was directly affected by the financial crisis. Applying a difference-in-differences approach, they found no statistically significant differences in the provision of apprenticeship training between firms that were affected by the financial crisis and firms that were not. However, there was a general decrease in firms’ willingness to participate in apprenticeship training (extensive margin), while the authors found no effects regarding training intensity (i.e., the share of apprentices in relation to all employees). Thus, the results of Bellmann et al. (2014) suggest that the crisis mainly affected smaller firms that, as a response, no longer offered apprenticeship positions after the start of the financial crisis. Muehlemann et al. (2020) analyzed the association between firms’ business cycle expectations (as measured by the Ifo Business Climate Index and the Ifo Employment Barometer) and the apprenticeship market based on data that included the universe of training contracts as well as the number of unfilled vacancies and unsuccessful applicants registered with the Federal Employment Agency. They applied first differences (FD) panel regressions and found that the financial crisis in 2008 reduced firm demand for apprentices in the following year by 6.5% and the number of apprenticeship contracts by 4.7%.17 Baldi et al. (2014) also analyzed the association of the business cycle and the number of apprenticeship contracts in Germany between 1999 and 2012, applying FD regression models. Based on a state-level panel regression, they found only a weak association between state-level output growth and new apprenticeship contracts during the period 1999–2007 but a positive association after the start of the financial crisis in 2008 for both East and West Germany.

For Switzerland, Lüthi and Wolter (2020a) analyzed a canton-level panel data set on the number of apprenticeship contracts for the period 1987–2016 and applied both FD and ADL panel regressions. They used the cantonal unemployment rate and the GDP per capita growth rate as business cycle measures. Regarding the Great Recession, they found that, based on the results of their ADL model, Swiss GDP fell substantially (by 2.1 percentage points), but the crisis led to only a 1% decrease in the number of apprenticeships in the long run. They also found that firms did not react differently during the Great Recession than during earlier recessions.18

For the United Kingdom, Felstead et al. (2013) provided descriptive statistics based on data from the Quarterly Labour Force Survey and reported that apprenticeships fell from 191,000 in the first quarter of 2009 to 140,000 one year later. They noted, however, that there had already been a negative trend in apprenticeship contracts since 2005; thus, similar to the experience in Germany, the entire decrease in apprenticeships could not be attributed to the financial crisis. In more recent work, however, Ventura (2020) reported the number of apprenticeships started based on individualized learner records and found only a small decrease in apprenticeships in 2009 in comparison to the level in 2008 for apprentices aged 16–24, while the number of positions for apprentices aged 25 and older actually increased.

Table 2. Institutional Settings in Apprenticeship Countries

Training duration

Training costs and benefits

Financial subsidies for training firms and training levies

Apprentice pay

Time spent in the workplace


2–3, 5 years

Average net training costs of 6,500 euros per apprentice and year of training (Schönfeld et al., 2020)

No direct subsidies for firms, training levy in construction industry and for nursing (only some states)

20–45% of skilled worker pay, sectoral/regional apprentice pay determined by CBAs; firms not covered by a CBA can pay max. 20% below tariff; national minimum apprentice wage introduced in 2020



4 years


Direct subsidies for firms (14,800 euros per apprenticeship position for two years), no training levy

30–80% of skilled worker pay, depending on year of training, determined at sectoral level



3–4 years, 2-year programs for less academically inclined individuals

Average net training benefit of 3,000 Swiss francs per apprentice and year of training (Gehret et al., 2019)

No direct subsidies, some regional training funds, possibility for government to introduce mandatory sectoral training funds

16% of skilled worker pay (Muehlemann et al., 2013), no minimum apprentice wage but wage recommendations in some sectors


United Kingdom

min. 12 months

Average net training costs between 1,200 and 7,200 GBP per apprentice and year of training (Gambin et al., 2010)

Training levy, additional 10% government support, grants for firms training apprentices aged 16–18

63% of skilled worker wage in metalworking industry

Up to 80%

Notes: CBA, collective bargaining agreement.

Source: Adapted from Brunello (2009); International Labour Organisation (2012); Kuczera (2017); and OECD (2018).

The COVID-19 Crisis and Apprenticeship Markets in Germany and Switzerland

This section discusses the effects of the COVID-19 pandemic on the apprenticeship market and focuses on Germany and Switzerland, two countries with a long tradition of apprenticeships as well as large apprenticeship systems in quantitative terms. Although the pandemic likely had adverse effects in other countries, the focus on these two countries is motivated by the fact that data on apprenticeship market outcomes are readily available and a series of empirical studies covering this particular time span have already been published.


In Germany, starting on March 16, 2020, several states started to close schools, retail businesses, and churches. On March 18, the German stock market index reached its lowest value since 2013, reflecting a strong change in the expectations of analysts. Even though the stock markets recovered from much of the loss experienced in the following weeks, it became clear that the German economy would be hit hard by the crisis. The German Institute for Vocational Education and Training (BIBB) predicted in May 2020 that the crisis would severely impact the German apprenticeship market, mainly through a reduction in demand for apprentices (Maier, 2020). The BIBB calculated different scenarios based on the assumptions of various expert groups about the decrease in German GDP in spring 2020, which ranged from −2.8% to −11.2%, underscoring the vast amount of uncertainty at that time. Muehlemann et al. (2020) took a different approach to predict the effects of the crisis, using survey data on business expectations up to June 2020. They predicted that COVID-19 would lead to an 8.1% decrease in firm demand for apprentices and a 6% decrease in the number of training positions,19 which roughly corresponded to a scenario in Maier (2020) that assumed a 7% decrease in annual GDP.

In response to the crisis in the apprenticeship market, the German government focused on direct subsidies to training firms to alleviate adverse effects.20 Training firms could apply for a premium of 2,000 euros for each completed training contract if they maintained training at the same level as the previous year and a premium of 3,000 euros for each additional training contract concluded beyond their precrisis training level. Moreover, companies that took on apprentices from firms that had to file for bankruptcy received a bonus of 3,000 euros per apprentice accepted. In addition, funding of 75% of the gross training allowance was provided to companies that did not apply for short-term work for trainees and their trainers.

In December 2020, the BIBB published statistics on the situation in the German apprenticeship market as of September 30, 2020 (Oeynhausen et al., 2020). While firm demand for apprenticeships decreased by 9.1% as compared with 2019, the number of apprenticeship contracts decreased by 11%. However, by the end of September, a record 70,000 training vacancies were still unfilled, and 60,000 individuals were still actively looking for apprenticeships. These figures indicated that the contact restrictions associated with COVID-19 likely had a negative impact on the matching process between supply and demand in the German apprenticeship market. However, the decrease in apprenticeships was particularly pronounced in tourism industry occupations, in hotels, and in restaurants, which were hit most severely by COVID-19 due to travel bans and lockdowns.

Figure 1. Development of the German apprenticeship market from January 2020 to September 2020.

Source: Federal Employment Agency (2020).

In terms of the monthly development since January 2020, the number of signed contracts during the first lockdown was lower than in the previous year because contact restrictions made it difficult for interested applicants and potential training firms to meet (see Figure 1). However, two developments are of particular interest. First, it is apparent that firms’ expectations about the business climate and the labor market decreased strongly starting in March 2020. In the following month, the number of posted vacancies and the number of registered applicants started to decrease in comparison to the numbers posted and registered in the same month in 2019, but the number of unsuccessful applicants started to increase. After the end of the lockdown in April, business expectations started to increase strongly from May through September, and the number of filled vacancies started to recover vis-à-vis their number in 2019, from −16% in June to −9% in September. However, by September 2020, severe matching problems emerged in the apprenticeship market, as both the number of unsuccessful applicants (+20%) and the number of unfilled registered vacancies (+13%) were higher than in 2019. Second, it is important to note that both the number of registered applicants and the number of apprenticeship vacancies in January and February, when few firms and businesses did not expect to be strongly affected by COVID-19, were approximately 5% lower than in the previous year. Thus, factors other than COVID-19 contributed to the lower number of training contracts in 2020, such as the decreasing number of school leavers in Germany, fewer individuals in the transitory system than in previous years, and the overall lower level of business expectations than in the previous year even before the first COVID-19 outbreaks in Germany.


In Switzerland, like Germany, a general lockdown was imposed by the federal government on March 16, 2020, with the closure of all businesses (except for those providing essential goods) and the obligation to work from home accompanied by border closures, which brought all tourism to a standstill. The economic consequences were evident in the first major revision of the GDP forecast by the State Secretariat for Economic Affairs (SECO), which in April predicted a 6.7% decline in GDP after positive growth for 2020 had been projected in December 2019.

Based on these forecasts and the model of Lüthi and Wolter (2020a), a decline in apprenticeship contracts of between −1.7% and −3.7% (95% confidence interval) would have been expected for 2020 in comparison to the number of contracts in 2019 (Lüthi & Wolter, 2020b). From April to June, the cumulative number of signed apprenticeship contracts developed roughly within this forecast interval (see Figure 2). In July, however, the number of contracts exceeded that in the forecast interval and ended up at the same value as in the previous year, despite the COVID-19 crisis. Even if it is impossible to conclusively state the exact reasons for this positive development, some factors can be identified and divided into four groups: first, short-term reactions in the search activities of apprenticeship seekers; second, changes in economic trends (GDP and unemployment); third, measures to support companies in the short run without directly influencing GDP; and fourth, measures to directly influence the apprenticeship market.

It cannot be completely ruled out that the rapid slump in the number of signed apprenticeship contracts in the wake of the shutdown in March might also be attributed to the fact that individual companies waited before signing apprenticeship contracts. However, the picture of the short-term slump and then the marked catch-up process after the lifting of most restrictions at the end of May 2020 is more consistent with the explanation regarding the search intensity of prospective apprentices. Based on the approximately 10 million search queries on the platform of the national apprenticeship exchange, Goller and Wolter (2021) showed that the temporal pattern of these search queries is practically identical to the pattern of signed apprenticeship contracts. The temporary fluctuations over the months—but not the fact that the number of apprenticeship contracts in the end did not decline despite the recession—can therefore be explained by the supply of apprentices. To account for the latter outcome, the other three explanations must be considered.

Figure 2. Development of the total number of signed apprenticeship contracts in Switzerland in 2020 (differences in the cumulative numbers of contracts in comparison to those in the same months in the previous year).

Data source: Task Force

In September, SECO reduced the forecasted GDP slump for 2020 to −5% and further revised the forecast in December 2020 to −3.3%. Even if this value may still be corrected, the December forecast would imply a corresponding change in the forecast interval for the apprenticeship market to between −0.8 and−2.2% (i.e., half of the improvement relative to the original forecast interval), because the economy developed much better than had been expected in the first weeks of the lockdown. The slightly better real development than the lower bound of the last model forecast would thus result in an estimation error that would no longer be of great significance. This development also clearly shows that forecasts of the development of the apprenticeship market, which is itself based on forecasts, namely, of GDP and the unemployment rate, are very susceptible to errors in economic forecasting. Whether the use of expectation data from firms (as in Muehlemann et al., 2020) would have produced a better forecast from the outset is questionable. At the first moment of the lockdown, the firm expectation data from the Economic Sentiment Indicator, for example, also showed a historically sharp slump at the beginning that then recovered steadily over the course of the year.

As explanations of the now approximately 1.5% more apprenticeship contracts in 2020 than a model based on the experience of previous economic cycles would have predicted, the two groups of policy measures are now also considered. Economic measures that did not primarily serve to support economic growth included, for example, a temporary ban on debt collection and the generous granting of interest-free loans. This also made it possible to maintain the supply of apprenticeships in firms that had already planned to continue offering training places and were counting on a short and temporary economic dip that would be over more or less immediately after the end of the lockdown. Thus, effects on the demand for apprentices from bankruptcies due to the lockdown were forestalled. Finally, the measures that directly targeted the apprenticeship market included those aimed at both firms (such as financial incentives from the cantons) and apprenticeship seekers. For example, the period in which apprenticeship contracts could still be concluded for 2020 was extended from August to the beginning of November. The latter measure primarily served to give young people who were limited in their apprenticeship search during the lockdown more time to find a suitable training firm. Additionally, several projects were supported that were intended to improve the overall matching of supply and demand so that open apprenticeship positions would not remain unfilled (e.g., online matching platforms). It will never be possible to quantify the effect of these measures, but their overall impact can be estimated at approximately 1.5% of apprenticeships—assuming that the forecast model correctly estimated the influence of the economy on the apprenticeship market in 2020 as well.

Lessons Learned

The effects of COVID-19 differ strongly between Germany and Switzerland. In Switzerland, the number of apprenticeship contracts concluded in 2020 during the crisis turned out to be somewhat higher than could have been expected in view of the economic developments. However, previous correlations established between the economy and the apprenticeship market cannot be called into question because of this: On the one hand, the forecast error of a few hundred apprenticeship placements for a total of over 70,000 apprenticeship contracts is too small, and on the other hand, this error is still based on the forecast of economic development. In Germany, both the demand for apprentices and the number of apprenticeship contracts decreased substantially due to COVID-19, as was initially predicted based on the observed association between apprenticeships and business cycle expectations. These findings imply that the German economy and its apprenticeship market reacted more strongly to this crisis than those of Switzerland.

The comparison of Switzerland and Germany in the current crisis and in the Great Recession shows that both the macroeconomic effects and the reaction of the apprenticeship market to the decline in GDP were larger in Germany.21 However, it is difficult to identify the potential reasons for the differences in business cycle effects, such as differences in policy measures or the fact that Swiss firms on average generate a net training benefit while most German firms have investment-oriented training motives. Based on the theoretical considerations, however, this observation supports the view that firms that train for investment-oriented reasons suffer more from strong economic downturns. The expected post-training benefits decrease because firms expect to fill fewer skilled worker vacancies in the future. Conversely, the financial risk of taking on new apprentices is substantially lower for firms with a production-oriented training motive, as is the case for most Swiss training firms.


Business cycle fluctuations affect not only labor markets but also apprenticeship markets. Empirical studies consistently find a positive association between the business cycle, firm demand for apprentices, and the number of apprenticeship contracts. However, the degree to which an apprenticeship market is affected by macroeconomic shocks differs across countries. Empirical studies analyzing Swiss apprenticeships have found that business cycle fluctuations, although statistically significant, had an economically small effect on the number of training contracts concluded during the Great Recession and have been almost fully absorbed in the current COVID-19 crisis. Conversely, in Germany, the Great Recession led to an almost 5% decrease in apprenticeships, and the economic effects related to COVID-19 strongly reduced the number of apprenticeship positions in 2020.

The existing empirical evidence suggests that severe mismatch problems arose during the COVID-19 pandemic, as it became more difficult for training firms and potential apprentices to meet. Moreover, information asymmetries about suitable training occupations and vacancies for interested individuals increased as many training fairs did not take place as planned, and in-person visits of potential training firms either could not take place at all or had to be moved online. Although the empirical evidence does not allow a causal interpretation, the difference in the outcomes of the German and Swiss apprenticeship systems is in line with theoretical considerations. German training firms on average need to make a substantial net investment in apprenticeship training—a financial risk that some firms were no longer willing (or able) to take during an economic downturn as their expected number of future skilled worker vacancies decreased. Thus, framework conditions allowing firms to train apprentices cost-effectively may be helpful in stabilizing apprenticeship markets over the business cycle. However, the current empirical evidence does not allow us to draw very strong conclusions regarding this issue, particularly because the external validity of our findings is limited by the evidence, which comes from only two countries.

Future research should focus on analyzing business cycle effects in other countries with apprenticeship systems and on the possible heterogeneity in business cycle effects across economic sectors and training occupations. Currently, however, most countries lack detailed and representative data on the demand for and the supply of apprentices across occupations and industries, which would allow us to obtain an improved understanding of how business cycle fluctuations affect apprenticeships and what measures are effective in stabilizing apprenticeship markets during times of economic crisis under different regimes. Governments should therefore try to establish or improve the necessary statistics. Moreover, there is a paucity of research that focuses on the retention behavior of training firms over the business cycle. Such knowledge would be particularly important to better understand the training behavior of investment-oriented training firms, as they need to generate post-training benefits to recoup their initial training investment. Administrative establishment-level data linked with individual information about the characteristics of an apprentice would allow research to go one step further, as heterogeneity at the firm level (in terms of production processes, technology adaptation, management strategy, and culture) may in part determine a firm’s reaction to business cycle fluctuations. Finally, government initiatives vary not only between countries but also across regions within a particular country. Such variation could potentially be exploited to evaluate the effectiveness of measures taken during economic downturns that are aimed at stabilizing apprenticeship markets.


  • Acemoglu, D., & Restrepo, P. (2019). Automation and new tasks: How technology displaces and reinstates labor. Journal of Economic Perspectives, 33(2), 3–30.
  • Askilden, J. E., & Nilsen, Ø. A. (2005). Apprentices and young workers: A study of the Norwegian youth labour market. Scottish Journal of Political Economy, 52(1), 1–17.
  • Baldi, G., Brüggemann-Borck, I., & Schlaak, T. (2014). The effect of the business cycle on apprenticeship training: Evidence from Germany. Munich Personal RePEc Archive Paper, 59(116), 1–15.
  • Bellmann, L., Gerner, H.-D., & Leber, U. (2014). Firm-provided training during the Great Recession. Jahrbücher für Nationalökonomie und Statistik, 234(1), 5–22.
  • Blatter, M., Muehlemann, S., & Schenker, S. (2012). The cost of hiring skilled workers. European Economic Review, 56(1), 20–35.
  • Blatter, M., Muehlemann, S., Schenker, S., & Wolter, S. C. (2016). Hiring costs for skilled workers and the supply of firm-provided training. Oxford Economic Papers, 68(1), 238–257.
  • Brunello, G. (2009). The effect of economic downturns on apprenticeships and initial workplace training: A review of the evidence. Empirical Research in Vocational Education and Training, 1(2), 145–171.
  • Chow, G. C. (1989). Rational versus adaptive expectations in present value models. Review of Economics and Statistics, 71(3), 376–384.
  • Coibion, O., & Gorodnichenko, Y. (2012). What can survey forecasts tell us about information rigidities? Journal of Political Economy, 120(1), 116–159.
  • Coibion, O., & Gorodnichenko, Y. (2015). Information rigidity and the expectations formation process: A simple framework and new facts. American Economic Review, 105(8), 2644–2678.
  • Dietrich, H., & Gerner, H.-D. (2007). The determinants of apprenticeship training with particular reference to business expectations. Zeitschrift für Arbeitsmarktforschung, 40(2/3), 221–233.
  • Federal Employment Agency. (2020). Ausbildungsstellenmarkt.
  • Federal Ministry of Labour and Social Affairs. (2013). Begleitforschung “Auswirkungen des Ausbildungsbonus auf den Ausbildungsmarkt und die öffentlichen Haushalte” (Forschungsbericht Arbeitsmarkt Nr. 438).
  • Felstead, A., Green, F., & Jewson, N. (2013). The impact of the 2008–9 recession on the extent, form and patterns of training at work (LLAKES Research Paper 22). Centre for Learning and Life Chances in Knowledge Economies and Societies.
  • Gambin, L., Hasluck, C., & Hogarth, T. (2010). Recouping the costs of apprenticeship training: Employer case study evidence from England. Empirical Research in Vocational Education and Training, 2(2), 127–146.
  • Gehret, A., Aepli, M., Kuhn, A., & Schweri, J. (2019). Lohnt sich die Lehrlingsausbildung für die Betriebe: Resultate der vierten Kosten-Nutzen-Erhebung. Eidgenössisches Hochschulinstitut für Berufsbildung.
  • Gehrke, B., Lechthaler, W., & Merkl, C. (2019). The German labor market during the Great Recession: Shocks and institutions. Economic Modelling, 78, 192–208.
  • Goller, D., & Wolter, S. C. (2021). “Too shocked to search”: The COVID-19 shutdown’s impact on the search for apprenticeships. Swiss Journal of Economics and Statistics.
  • Hanushek, E. A., Schwerdt, G., Woessmann, L., & Zhang, L. (2017). General education, vocational education, and labor-market outcomes over the lifecycle. Journal of Human Resources, 52(1), 58–87.
  • Hinz, T. (2019). Personnel policy adjustments when apprentice positions are unfilled. International Journal of Manpower, 40(5), 958–978.
  • Kuczera, M. (2017). Striking the right balance: Costs and benefits of apprenticeship (Education Working Paper No. 153). Organisation for Economic Co-operation and Development.
  • Lovell, M. C. (1986). Tests of the rational expectations hypothesis. American Economic Review, 76(1), 110–124.
  • Lüthi, S., & Wolter, S. C. (2020a). Are apprenticeships business cycle proof. Swiss Journal of Economics and Statistics, 156(3), 1–11.
  • Lüthi, S., & Wolter, S. C. (2020b, May 8). The impact of the COVID-19 crisis on the provision of apprenticeships in Switzerland. Leading House Short Note, 1–4.
  • Maier, T. (2020). Auswirkungen der “Corona-Krise” auf die duale Berufsausbildung: Risiken, Konsequenzen und Handlungsnotwendigkeiten. Bundesinstitut für Berufsbildung.
  • Méndez, F., & Sepúlveda, F. (2012). The cyclicality of skill acquisition: Evidence from panel data. American Economic Journal: Macroeconomics, 4(3), 128–152.
  • Muehlemann, S., Pfann, G. A., Pfeifer, H., & Dietrich, H. (2020). Supply shocks in the market for apprenticeship training (Leading House Working Paper No. 143). University of Zurich.
  • Muehlemann, S., Pfeifer, H., Walden, G., Wenzelmann, F., & Wolter, S. C. (2010). The financing of apprenticeship training in the light of labor market regulations. Labour Economics, 17(5), 799–809.
  • Muehlemann, S., Pfeifer, H., & Wittek, B. H. (2020). The effect of business cycle expectations on the German apprenticeship market: Estimating the impact of Covid-19. Empirical Research in Vocational Education and Training, 12(8), 1–30.
  • Muehlemann, S., Ryan, P., & Wolter, S. C. (2013). Monopsony power, pay structure, and training. ILR Review, 66(5), 1097–1114.
  • Muehlemann, S., & Strupler Leiser, M. (2018). Hiring costs and labour market tightness. Labour Economics, 52(C), 122–131.
  • Muehlemann, S., & Wolter, S. C. (2014). Return on investment of apprenticeship systems for enterprises: Evidence from cost-benefit analyses. IZA Journal of Labor Policy, 3(25), 1–22.
  • Muehlemann, S., & Wolter, S. C. (2020). The economics of vocational training. In S. Bradley & C. Green (Eds.),The economics of education: A comprehensive overview (2nd ed., pp. 543–554). Elsevier.
  • Muehlemann, S., Wolter, S. C., & Wüest, A. (2009). Apprenticeship training and the business cycle. Empirical Research in Vocational Education and Training, 2, 173–186.
  • Organisation for Economic Co-operation and Development. (1999). Preparing youth for the 21st century: The transition from education to the labour market.
  • Organisation for Economic Co-operation and Development. (2010). Learning for jobs.
  • Organisation for Economic Co-operation and Development. (2020). OECD indicators on employment protection.
  • Oeynhausen, S., Milde, B., Ulrich, J. G., Flemming, S., & Granath, R. O. (2020). Die Entwicklung des Ausbildungsmarktes im Jahr 2020. Federal Institute for Vocational Education and Training.
  • Pissarides, C. A. (2000). Equilibrium unemployment theory (2nd ed.). MIT Press.
  • Sauer, S., & Wohlrabe, K. (2020). Ifo Handbuch der Konjunkturumfrage. Ifo Beiträge zur Wirtschaftsforschung, 88, 1–362.
  • Schlögl, P., Mayerl, M., Löffler, R., & Schmölz, A. (2020). Supra-company apprenticeship training in Austria: A synopsis of empirical findings on a possibly early phase of a new pillar within VET. Empirical Research in Vocational Education and Training, 12(1), 1–17.
  • Schönfeld, G., Wenzelmann, F., Pfeifer, H., Risius, P., & Wehner, C. (2020). Ausbildung in Deutschland–eine Investition gegen den Fachkräftemangel (Ergebnisse der BIBB-Kosten-Nutzen-Erhebung 2017/18. BIBB Report 1/2020).
  • Schweri, J., Kuhn, A., & Wolter, S. C. (2019). Local norms describing the role of the state and the private provision of training (Working Paper No. 7519). Center for Economic Studies.
  • Schweri, J., & Müller, B. (2008). Die Ausbildungsbereitschaft der Betriebe: Entwicklungen 1995 bis 2005. Bundesamt für Statistik, 18(1), 359–375.
  • Stevens, M. (1999). Human capital theory and UK vocational training policy. Oxford Review of Economic Policy, 15, 16–32.
  • Ventura, G. (2020). What future for apprenticeships after coronavirus? (Briefing Note No. 012). London School of Economics and Political Science.
  • Wolter, S. C., & Ryan, P. (2011). Apprenticeship. In E. A. Hanushek, S. Machin, & L. Woessmann (Eds.), Handbook of the economics of education (Vol. 3, pp. 521–576). North-Holland.


  • 1. To the extent that part-time training instructors have less work (i.e., do not work at full capacity) during a recession, however, the opportunity costs of providing training decrease as well.

  • 2. In addition to the state, there are other actors that could intervene in the apprenticeship market, such as employer associations or nongovernmental organizations. For reasons of space, however, only the state is discussed in detail here.

  • 3. An example is the case of supracompany apprenticeship training in Austria (Schlögl et al., 2020).

  • 4. The German employment agency, for example, reports the number of unfilled training positions as the difference between all registered apprenticeship vacancies and the number of successfully filled vacancies. However, these statistics do not include apprenticeship positions that were filled or remained unfilled but that were not previously registered with the employment agency. In recent years, however, the share of registered vacancies in relation to all signed apprenticeship positions has increased substantially (Maier, 2020). As an alternative, representative establishment-level data can provide information about unfilled vacancies (Hinz, 2019), although cell sizes typically become very small at the regional level, particularly for less popular training occupations.

  • 5. The German employment agency collects data on the number of registered individuals who showed interest in an apprenticeship position at some point during a particular year (Ausbildungsinteressierte).

  • 6. Germany and Switzerland usually report the number of young individuals who end up in the transitory system. Germany also reports monthly statistics on the number of registered individuals who are actively looking for an apprenticeship and whether they have already found an alternative education option (Federal Employment Agency, 2020).

  • 7. Conversely, data based on the universe of training contracts allow for a fine-grained analysis at the occupational and local levels.

  • 8. The responsiveness of unemployment to business cycle fluctuations, however, may differ strongly across countries because of differences in employment protection legislation and labor market policies. For example, unemployment increased strongly in the United States after the Great Recession, whereas unemployment hardly increased at all in Germany, arguably due to labor market reforms before and short-time work programs during the crises (Gehrke et al., 2019). Moreover, hiring and firing costs, unionization, and the presence of works councils may vary across industries within a particular country and thus lead to different responses in terms of a firm’s hiring and layoff decisions.

  • 9. The training decision differs in the case of continuous vocational education and training for existing employees that lasts only a few days or week. In such a scenario, the provision of training may be counter-cyclical, as the opportunity costs of training are lower during a recession (cf. Brunello, 2009).

  • 10. Note that the BCI incorporates both a firm’s estimates about the current situation and its expectations for the next 6 months.

  • 11. However, while full-information rational expectations (FIRE) are often assumed in macroeconomic models, the empirical evidence questions the validity of this assumption (e.g., Chow, 1989; Lovell, 1986). More recent research confirms that expectations deviate substantially from FIRE and highlights the role of information rigidities as well as important heterogeneities in microdata (cf. Coibion & Gorodnichenko, 2012, 2015).

  • 12. Note that macroeconomic shocks do not affect labor markets immediately because in many countries firms cannot fire employees immediately. To the extent that unemployment is used as a proxy for current business expectations in period t, the empirical model should include unemployment in period t + 1. Moreover, to the extent that adjustment costs (layoff costs and future hiring costs) are sizable, firms will not lay off employees, depending on the expected duration of the economic downturn.

  • 13. To the extent that a firm’s expectations are systematically biased, however, the results based on actual GDP and labor market outcomes are expected to deviate from results based on business expectations.

  • 14. Some studies are conducted at the firm level and can thus also include firm-level fixed effects (Dietrich & Gerner, 2007).

  • 15. In Germany, the number of school leavers started to decrease strongly beginning in 2007. While the financial crisis indeed had an adverse effect on the apprenticeship market, simply looking at the change in apprenticeship contracts before and after the financial crisis would lead to overestimation of the effects of the business cycle (cf. Muehlemann et al., 2020). Similarly, the United Kingdom was already experiencing a negative trend in its number of apprenticeship contracts prior to the financial crisis (Felstead et al., 2013).

  • 16. Note that the difference in the cycle effects between the two estimation models is not statistically significant.

  • 17. However, they find no statistically significant association between business expectations and unsuccessful applicants for the period 2007–2019.

  • 18. Lüthi and Wolter (2020a) found no evidence for asymmetric effects over the business cycle (i.e., that effects are different during downturns from those during boom periods), which is in line with earlier results for Germany (Dietrich & Gerner, 2007).

  • 19. These effects are broadly in line with those in Dietrich and Gerner (2007), who examined the effects of business expectations at the firm level during the period 1993–2003. They found that a 1% drop in business expectations is associated with a 0.35–0.4% decrease in the growth of training positions. As the Ifo Business Climate Index dropped by 13 points in the first two quarters of 2020 in comparison to the level in the previous year (when the index was at 99), the results of Gerner and Dietrich would imply a reduction in the number of training contracts of approximately 5%.

  • 20. See

  • 21. During the COVID-19 crisis, the decline in GDP for Germany was expected to be −5% in comparison to Switzerland’s −3.3%. During the Great Recession, GDP fell by −5.7% from 2008 to 2009 in Germany, while in Switzerland, GDP decreased by only −2.1%.