THE DETERMINANTS OF YOUTH TEMPORARY EMPLOYMENT IN THE ENLARGED EUROPE: Do labour market institutions matter?

Temporary employment contracts have become a subject of fierce discussion in both scientific literature and public debate. While proponents claim that temporary contracts increase labour market flexibility by reducing the firing costs for employers, opponents argue that temporary contracts are associated with poor working conditions. Many empirical studies have found evidence for this latter ‘segmentation’ perspective in Western Europe (e.g., Giesecke and Groß 2004). In this regard, DiPrete et al. (2006) interpret temporary employment as a crucial new social inequality in Europe.

Given the disadvantages of temporary employment, the question of who is most at risk of such insecure and inferior employment becomes important. Whereas the core workforce is still relatively well sheltered in most European countries, prominent comparative studies reveal that youth are most at risk of needing to accept flexible employment forms (Schömann et al. 1998; Blossfeld et al. 2005). While we expect to confirm the previous findings of high relative incidence of temporary employment among youth, the central question of this paper is, however, how the youth relative temporary employment shares vary between countries and whether this variation can be explained by country-specific labour market institutions.

Previous research has shown that, apart from macroeconomic factors, the country-specific institutional setting may mediate individual-level relations in social stratification processes and account for the enormous cross-country variations (Mayer 2004; Hout and DiPrete 2006). For example, Esping-Andersen (2000) and Breen (2005) reveal that the cross-country variation in youth to adult unemployment can be explained with macro-structural and labour market institutional factors, such as employment protection legislation and union strength. Our article ties in with this research by analyzing the age-divide in temporary employment patterns.

A quantitative assessment of this intervening link for youths' exposure to temporary employment is still scarce. Maurin and Postel-Vinay (2005) and Polavieja (2006) investigate for Western Europe the impact of macro factors on temporary employment rates in general without having a specific focus on youth. Polavieja's (2006) analysis reveals that particularly employment protection rules, the degree of coordinated centralization of the collective bargaining system and the overall unemployment rate are the central determinants of overall temporary employment rates. Kahn (2007) uses a highly selective sample of eight countries using IALS data 1994–1998. He finds that strict employment protection for regular jobs and higher collective bargaining coverage raise the relative incidence of temporary employment for young workers.

Our study will re-examine these findings by advancing the analysis from three viewpoints. First, we extend the comparison to a larger set of countries using comparable data from the European Labour Force Survey (EULFS) 2004. It includes datasets from Central and Eastern European (CEE) societies, whose experience has been ignored in previous studies. The post-socialist countries, with their specific institutional arrangements and structural conditions, can serve as valuable basis for cross-country comparisons of individual risk patterns in Europe. Furthermore, a larger number of macro cases should improve the quality of our analysis and increase the coverage of our results. Second, we employ multi-level regression techniques in order to disentangle the roles of individual characteristics on the one hand, and the country-level attributes, on the other (Snijders and Bosker 1999). In contrast to standard regression analysis used in previous studies, multi-level models allow for unobserved heterogeneity both at the micro and macro level. Finally, apart from a detailed and theory-driven evaluation of the influence of specific labour market institutions (i.e., employment protection legislation and Unions' bargaining power) on the youth relative probabilities of having a temporary contract, we control for potentially confounding macro-structural factors. Specifically, we investigate the role of economic conditions such as the overall unemployment rate and business uncertainty, which might also explain the cross-country variation.

This paper is organised as follows: Section 2 addresses potential influences of labour market institutions on youths' relative temporary employment probabilities. Based on these theoretical arguments, we derive hypotheses for the empirical analysis. Section 3 describes the data set, variables, and the statistical methods used. Section 4 presents descriptive evidence and the results of the multi-level analysis. Finally, Section 5 concludes the paper.

The existing empirical studies reveal that in many countries particularly youth are hit by employment flexibilisation because they lack experience and seniority (Blossfeld et al. 2005; Kahn 2007). We expect to find this pattern of the relatively higher incidence of temporary employment among young people in all European countries. However, there are well established theoretical approaches which motivate hypothesis on high cross-country heterogeneity in youth relative temporary employment rates (Hypothesis 1).

This variation might be related to micro-level factors. Previous comparative studies for Europe have shown that the risk of holding a temporary contract depends on individual as well as job-related characteristics (Schömann et al. 1998; Van der Velden and Wolbers 2003; Giesecke and Groß 2004; Polavieja 2006). Individual characteristics such as lower education or being female increase the probability of having a temporary contract. But job characteristics such as the firm size and industry are also important determinants in this respect. Therefore it is necessary to take the micro-level determinants into account when trying to explain cross-country differences in youth relative temporary employment rates.

At the macro-level, the age-specific temporary employment probabilities may be mediated by specific national institutional arrangements (Bertola et al. 2007; Kahn 2007). We focus on the institutional settings of the labour market, which can affect the risks of getting a temporary contract. In particular, the roles of employment protection legislation and union strength are considered.

While the objective of employment protection legislation is to reduce the exposure of workers to unfair actions, these regulations may also increase employer costs for hiring and firing workers. Firing regular employees with permanent contracts may become very expensive due to high direct costs such as severance payments and indirect costs related to procedural difficulties (Bentolila and Bertola 1990). Strict employment protection legislation encourages employers to hire workers for temporary contracts, which eliminate potential firing costs because they end automatically after their expiration. Thus, we should observe a positive relationship between the protection of regular jobs and the overall incidence of temporary jobs.

In this respect, Kahn (2007) argues and finds evidence for a small sample of Western European countries that this effect will disproportionally hit young workers: if firing costs for permanent jobs are substantial, employers will be relatively reluctant to hire young workers into such jobs because their productivity is rather unknown due to their limited work history, which increases the risk of a poor job–worker match. Furthermore, younger workers have had fewer opportunities to convert their temporary jobs into permanent ones due to their shorter employment experience. Hence, more stringent employment protection for regular jobs is predicted to increase particularly the relative incidence of temporary employment for young workers (Hypothesis 2).

Besides regulations of permanent contracts, one can expect that the regulation of temporary employment also affects the risk of having a temporary contract. For example, Nunziata and Staffaloni (2007) conclude from their theoretical models that lower constraints on the use of temporary work increase the likelihood of having a temporary job. However, empirical evidence indicates that temporary employment regulation has no effect on the incidences of temporary employment, controlling for the strictness of employment protection for regular workers (Booth et al. 2002). Instead, it is argued that the recent increase of temporary employment in some Western European countries is due to partial deregulation, i.e., the combination of high employment security for permanent contracts and low barriers on the use of temporary contracts (Esping-Andersen 2000). Esping-Andersen (2000) expects that the combination of high protection of permanent workers and weak regulations on fixed-term contracts strengthens the age divide in the labour market because it particularly hits youths. Therefore, we assume that the positive impact of strict permanent employment protection on temporary employment risk will be especially pronounced in countries with less restrictive temporary contract regulations (Hypothesis 3).

Unions' power in negotiating wages and employment conditions is another labour market institution that might shape the distribution of individual temporary contract risks (Polavieja 2006; Kahn 2007). According to insider–outsider theory, unions represent collective interests of labour market insiders, who have already gained permanent employment relationships in the primary labour market segment (Lindbeck and Snower 1989). Prominent comparative studies confirm that while the group of insiders largely consist of prime-age males, outsiders tend to be overrepresented among young people (Van der Velden and Wolbers 2003; Blossfeld et al. 2005; Bertola et al. 2007). The age-determined insider–outsider distinction can result from seniority principles promoted by social partners (Grossman 1983; Bertola et al. 2007). If unions attach relatively greater weight to prime-age workers, interests, they favour higher wage floors, which are above the level of the wage that could be offered to less experienced – and therefore less productive – young workers. Their actions increase the welfare of ‘core workers’ but have the side effect of hindering labour market integration of youth (Van der Velden and Wolbers 2003). Therefore, we conclude that the stronger the representation of insiders' interests in wage setting, the lower the relative chance for young people to attain permanent contracts (Hypothesis 4).

Finally, the general argument concerning increased temporary work risk through strict regulations limiting labour turnover is only true if low wages cannot compensate employers for high firing costs (Lazear 1990). The mechanism of compensating the high firing costs with lower labour costs can be prevented by unions, which are the central labour market institution that affects wage setting. If unions compress wages by setting a high wage floor, then collective bargaining may accentuate the effects of employment protection in barring younger workers from attaining permanent jobs (Kahn 2007). Furthermore, unions may monitor the compliance with employment protection regulation. This might apply if such regulations are written into law but often ignored by employers. Hence, we expect a positive interaction effect between regulation of permanent work and collective bargaining coverage (Hypothesis 5).

In order to test these hypotheses, we use data from 23 countries drawn from the European Labour Force Survey (EULFS), covering the second quarter of 2004. This database provides standardised, cross-sectional information on individuals compiled from national labour force surveys. The survey is designed to be representative of the working-age population in Europe, thus providing a unique database of large-scale, comparable surveys of labour market behaviour and employment issues in EU countries (Eurostat 2007). We constrain the sample to employees aged 15–64 who no longer participate in education, i.e., we exclude students and apprentices.

The central variable defining the type of employment contract is a binary indicator, coded 1 for temporary contracts and 0 for permanent contracts. Temporary employment is characterised by the agreement between employer and employee on objective conditions under which a job ends, such as a specific date or the completion of a task. In particular, this applies to occasional, casual or seasonal workers, temporary agency workers, workers on probationary periods, and workers under contract for a specific task (Eurostat 2007).

The set of explanatory variables reflecting the individual determinants of employment contract contains personal demographic and employment characteristics, as well as firm characteristics. Standard demographic variables include gender and age. Gender is dummy-coded, whereas age is grouped in intervals (15–24, 25–34, 35–54, and 55–64 years). To control for differences in educational attainment, we introduce the level of education. Highest level of education achieved is measured in terms of an augmented ISCED classification that distinguishes three levels of qualifications: compulsory education (ISCED 0–2); upper secondary and post-secondary, non-tertiary education (ISCED 3–4); and tertiary education (ISCED 5–6). We also control for firm characteristics in the form of firm size and industry. Firm size is differentiated into three groups: small firms (1–10 employees), medium-sized firms (11–50 employees), and large firms (more than 50 employees). Economic sector is measured according to nine aggregated NACE classifications. Firm characteristics should capture compositional differences in the structure of the economy.

In order to investigate the extent that cross-national variation in age-specific risk patterns of temporary employment could be explained by the specific institutional arrangements, we focus on two dimensions: employment protection legislation for temporary and permanent contracts, and union power in wage setting mechanisms. Employment protection legislation (EPL) influences employer cost of hiring and firing workers. It is measured with internationally comparable indices in line with OECD (2004), which consider the legislation on permanent employment and temporary employment.¹ Regular employment legislation determines the rules for hiring and firing permanent workers, notification requirements, and severance payments. Temporary employment legislation regulates the use of temporary contracts, their renewal and maximum duration, as well as the functioning of temporary work agencies. The higher the value of the EPL index, the stricter the employment protection legislation is. Table 1 reports the values for Eastern and Western European countries.

Regulation concerning regular employment varies strongly across countries. In the Czech Republic and Slovakia, the legislation is rather rigid, whereas in Hungary, the labour law imposes very few restrictions on regular dismissals. Of the Western countries, Portugal has the most restrictive regular employment protection. Regulations on temporary work legislation are much weaker in most CEE countries than in most Western European countries, with the exception of Lithuania, Slovenia, and Latvia. Temporary employment regulation also varies in Western Europe, where France and Spain are the most restrictive and Ireland, The Netherlands, and the United Kingdom the least restrictive.

Unions' strength in fostering wage policies in favour of the core workforce can be assessed by different indicators. To guarantee the comparability to previous comparative studies of temporary employment risks, we choose collective bargaining coverage that measures the percentage of all salaried workers (unionised and non-unionised) who are covered by collective agreements. Table 1 shows that collective bargaining coverage varies considerably, from 100 percent coverage in Slovenia to 15 percent coverage in Lithuania. Compared to Western Europe, collective bargaining coverage is relatively weak in most transition countries (Crowley 2004).

To account for macro-structural influences, we control for the unemployment rate and business uncertainty.² Holmlund and Storrie (2002) show that, in the event of a depressed labour market, workers are more likely to accept temporary jobs in order to escape the risk of unemployment. Table 1 indicates that there is considerable variation in unemployment rates, both among CEE countries and in Western Europe. Furthermore, under high economic uncertainty, firms tend to refrain from long-lasting commitments and may use flexible working arrangements to adapt to volatile markets (Breen 1997). We proxy the level of business uncertainty by the Economic Sentiment Indicator (ESI) constructed on the basis of compiled and harmonized data from business and consumer surveys (for details, see European Commission 2007). The ESI summarizes judgements and anticipations concerning diverse facets of economic activity in different sectors of the economy. Values greater than 100 indicate an above-average economic sentiment, i.e., a lower business uncertainty, whereas values below 100 indicate a below-average position. While business prospects were rather good in most CEE countries, with the exception of Poland, there are some Western European countries with high business uncertainty, such as the Netherlands, Germany, and Portugal.³

Descriptive evidence reveals a high variation in the extent of flexibilisation of employment relationships in Western, Central, and Eastern Europe (see Figure 1). Young people have a substantially higher risk than prime-age employees in all European societies but a considerable cross-country variation in the share of young people performing temporary work is also apparent, which clearly supports our Hypothesis 1. In general, one can observe a higher relative probability for youth in most Western and Northern European countries, while the relative probability is much lower in most Central and Eastern European as well as Southern European countries. However, there are also outliers like Poland and especially Slovenia with rather high concentration of temporary jobs among young people and the United Kingdom with very low relative risks.

In order to illustrate the associations between the national labour market institutions and the youth relative temporary employment risk, Figure 2 plots for each country the temporary employment ratio of youth compared to the prime age group against the EPL index for regular employment. The first graphical inspection shows that, contrary to Hypothesis 2, there is no clear linear relationship between the impact of young age on the risk of temporary work and regulation of permanent contracts. Although many countries have a similar degree of moderate employment protection, we find a huge variation in the temporary employment ratio for youth compared to prime-age workers. In some Southern, Central and Eastern European countries with moderate or high protection for permanent jobs, such as Portugal, Spain, Lithuania and the Slovak Republic, we can observe that, despite strictness of legislation, the temporary employment ratio for youth compared to prime-age workers is rather low. Thus, even if terminating permanent work contracts becomes very expensive due to strict employment protection, employers do not use temporary contracts for hiring young workers more frequently.

Looking at the union's role in negotiating wages, we find that it seems to be an important determinant of the youth relative temporary employment risk (see Figure 3), which supports Hypothesis 4. We can observe that in those countries, in which the role of social partners is not pronounced, the relative risk of young people having temporary contracts is relatively low. This is especially the case in Central and Eastern Europe but applies also to the United Kingdom and Ireland. On the other hand, in many Western and Northern European countries with high levels of collective bargaining coverage, young people face very pronounced levels of relative risk of temporary employment compared with prime-age workers. This pattern can be also observed in Slovenia, a country with the highest share of collective bargaining coverage and the highest relative risk for youth.

Though offering a general impression of the relative incidence of temporary employment of young people compared to prime-age workers in Europe and its bivariate association with central labour market institutions, the descriptive figures might be somewhat misleading because they ignore other confounding factors. On the one hand, individual characteristics like gender, education or job attributes might influence the temporary employment incidence; on the other hand, other country characteristics might explain the cross-country variations. Hence, the results of multivariate multilevel analyses are presented in this section (see Tables 2 and 3).

In the baseline model 1, we examine the impact of micro-level determinants of temporary employment, while allowing systematic cross-country variation in the intercept and slope coefficient of young workers. We can confirm our previous descriptive evidence in our large sample of European countries: young workers aged between 15 and 24 are more likely to find themselves in temporary jobs than the reference group of prime-age workers, even taking other individual characteristics and job attributes into account. This confirms our descriptive evidence.⁶

The other individual level variables mostly have the expected effects in accordance with previous studies (Schömann et al. 1998; Maurin and Postel-Vinay 2005; Polavieja 2006). For example, we find that workers with compulsory education have a lower chance of being employed with permanent contracts compared to workers who have upper secondary or tertiary education. Thus, higher educated and hence more productive workers have better and more stable employment conditions than the least-educated ones. Interestingly, tertiary educated workers do not have any advantage compared to those from upper secondary schools, which shows that there is no clear linear relationship between education and temporary employment. Women are more often employed with temporary contracts than men, which might be explained by their general labour market disadvantages and unstable employment careers. Enterprise characteristics, such as firm size and industry, also matter in the micro-level regression. For example, the larger the firm, the lower the probability of being temporarily employed. The temporary employment patterns are also structured across the industry sectors with relatively high prevalence in agriculture, construction, hotels and restaurants and service sectors. These are mainly sectors that are exposed to quick demand changes, which require a higher share of temporary workers as a buffer stock.

In sum, we can confirm the social risk patterns of temporary employment that have been found in the existing literature: young persons, lower educated workers, women, employees in small firms and cyclical industry sectors face higher risks of holding a temporary contract when compared to other social groups. Especially young workers are exposed to temporary employment and the level 2 variance component reveals that there is substantial cross-country variation in the relative temporary risk for young workers – even after controlling for micro-level determinants – which again supports our Hypothesis 1. The question thus arises which country characteristics might be responsible for this variation.

To answer this question, the random intercept and random slope coefficient for young workers are explained by macro-level factors in the following models. Model 2 investigates how cross-country variation in temporary employment risk is related to the employment protection of permanent jobs. In accordance with the graphical insight, there is no significant positive relationship between EPL for regular employment and the temporary employment probability in general as well as the impact of young age on the temporary employment risk. Hence, we find no evidence supporting Hypothesis 2. Higher firing costs for permanent contracts do not induce employers to use fixed-term contracts more often. Model 3 shows that the results do not change substantially when economic conditions are controlled for by including the unemployment rate and the economic sentiment index. Interestingly, the effect of the economic sentiment index has the expected impact on the relative youth temporary employment risk, but has no significant influence on the overall temporary employment incidence of other age groups. Thus, in times of economic uncertainty, employers use temporary contracts especially for youth.

Model 4 tests for the interaction effect between regulation of permanent and temporary jobs. The interaction term is insignificant, i.e., we find no evidence for Hypothesis 3, which expects that the positive impact of strict permanent employment protection on temporary employment risk is especially pronounced in countries with less restrictive temporary contract regulations. Furthermore, the point estimate for the direct effect of temporary employment regulation displays the wrong sign. This might be interpreted – albeit with some reservations – by policy endogeneity: in the case of low temporary employment incidence, governments might not see reasons for strict temporary work legislation, whereas in the case of high temporary employment risks, governments strengthen the restrictions of temporary contracts.

In sum, we do not find any evidence for significant influences of EPL on the likelihood of having a temporary contract in general and for youth in particular, which contradicts previous findings (Booth et al. 2002; Kahn 2007). The discrepancy between our results and existing studies might be explained by our larger number of countries allowing for more robust tests of macro-institutional influences. Furthermore, while other studies relied solely on country-level regressions or did not use multi-level techniques, we use multilevel models that allow controlling for micro-level determinants as well as random influences both at the micro and macro level. Another explanation of our findings suggested by some studies (Eamets and Masso 2005) is that high firing costs, approximated by high EPL, are no incentive for flexible temporary work contracts because in some countries firms are able to find alternative ways to achieve flexibility.

In order to quantitatively confirm Hypothesis 4, which states that the union role in wage negotiations increases the relative risk of temporary work for young people, we depart again from the multi-level model containing only micro-level regressors (model 5, which corresponds to model 1) and introduce the collective bargaining coverage in model 6. In accordance with our hypothesis, the results indicate that the higher the share of contracts that are subject to collective bargaining, the higher the incidence of temporary employment among young people. Hence, we can confirm Kahn's (2007) findings for our larger set of countries and using multi-level models. Interestingly, the Unions' effect on the overall temporary employment risk is also positive and significant. Thus, predominately youth are exposed to higher temporary employment risks in countries with high bargaining coverage.

The result remains robust even when economic conditions are taken into account in model 6. As in model 3, the economic sentiment variable has the expected negative sign for the relative temporary employment risk of youth but it is insignificant in this specification. Whereas the unemployment rate increases the risk of temporary employment for prime-aged workers, it has no additional effect on young workers. Thus, structural economic conditions also play a role for the social distribution of temporary employment relationships, which complements previous findings on relationships between economic conditions and youth unemployment risks (Van der Velden and Wolbers 2003; Breen 2005).

Finally, we allow for interactions between employment protection and collective bargaining coverage because, according to Hypothesis 5, we expect a connection between binding wage floors and the effects of employment protection mandates. Model 7 shows that the interaction term has the expected positive sign, i.e., more stringent employment protection in regular jobs raises young persons' risk of getting a temporary contract substantially more in countries with higher levels of collective bargaining coverage. Thus, strict regulations increase the temporary employment risk only if wages cannot adjust downwards for higher firing costs because of insider power represented through unions. However, the effect is not significant.

Taken together, the macro-institutional and macro-structural variables in the final model manage to explain about 54 percent (1–(0.114/0.212)×100%) of the macro-level variance in the slope for young workers. Although our aim has not been to provide a full explanation of the cross-country variation in youth–adult temporary employment ratios, our results reveal that a parsimonious set of theoretically derived institutional and structural indicators can explain a remarkable part of heterogeneity between nations.

The aim of this article has been to understand whether and to what extent the enormous cross-country variation of youth exposure to temporary employment relationships in European societies can be related to specific institutional labour market settings. A central finding of our study is that labour market institutional factors, together with macro-structural market conditions, shape the levels of relative youth temporary employment. This reinforces the view that the national institutional context of the labour market, as well as macro-structural influences like unemployment rate and business uncertainty, structure the labour market integration of young people (Müller and Gangl 2003).

Our results have policy relevance and implications because they show which institutional factors in the labour market may play a role in protecting youth from insecure temporary jobs. While previous research has revealed that employment protection legislation affects the youth–adult unemployment ratio (Esping-Andersen 2000), we can show that it seems to be of no importance for the youth–adult temporary employment ratio. Neither the strong protection of regular employment nor its combination with relaxed rules for temporary work seems to have significant influence.

What matters instead is the degree of insider protection due to Unions' bargaining power. In countries where the core workforce (insiders) has a protected position due to strong unions, young workers will find themselves as newcomers (outsiders) in a disadvantaged position when trying to get secure jobs. In contrast, in countries with weak unions, such as most of the CEE countries, all age groups are equally affected by market forces. Obviously, this does not imply that unions actually press young workers into precarious jobs. From a theoretical point of view, the observed pattern is the unintended consequence of the Unions' focus on improving the working conditions of the core workforce (Lindbeck and Snower 1989; Van der Velden and Wolbers 2003; Bertola et al. 2007).

Furthermore, it remains an open research question whether the impact of Unions' strength on youth to adult temporary employment ratio is related to the substitution of regular contracts with temporary contracts or whether it is result of labour market integration of otherwise unemployed youth. There are well established theoretical concepts which assume that strong unions under cooperative relationships between corporate partners can generate economically viable institutional structures of youth labour market integration (Soskice 1999). For example, such collective efforts might include training policies that reduce youth unemployment and integrate youth into the labour market. Then, the Unions' effect on youth temporary employment risks can be considered as more positive because they are finally the result of integrating otherwise unemployed youth into temporary positions. A final answer about the correct view cannot be derived from our results. However, our results provide a first detailed picture and ask for a comprehensive study of Unions' simultaneous influence on youth relative temporary employment and youth unemployment risks.

Finally, we must address the limitations of our methodological approach relying on the validity of the quantitative indicator-based measurement of institutions. Although using macro-indicators is the basic principle of quantitative international comparative research and necessary to guarantee comparability to previous research in the field, there are clearly limitations in this approach. Indicators may not fully capture the institutional settings in some countries, which may happen if the set of countries increases and becomes heterogeneous. Moreover, although the strength of our data lies in the high coverage of countries, high degree of comparability and representativeness, there are clearly limitations due to its cross-sectional character. If comparable longitudinal data becomes available for such a large set of European societies, future research should benefit considerably from analyzing the dynamics of youth temporary employment.

Anna Baranowska is Researcher at the Institute for Structural Research and Ph.D. candidate at the Institute for Statistics and Demography at Warsaw School of Economics. Her current research interests include atypical employment forms, transition to adulthood and early labour market career.

Michael Gebel is Researcher at the Mannheim Centre for European Social Research (MZES), University of Mannheim. His research interests are focused on school-to-work transition, atypical employment, and international comparative social research. He has published recently on these issues in the European Sociological Review, the Journal for Labour Market Research and Zeitschrift für Soziologie.