Atypical Employment and Flexicurity

Hitherto, discussion of flexicurity has focused on normal employment (Normalarbeitsverh„ltnis), with atypical work receiving only cursory attention. This paper attempts to remedy this conceptual oversight by identifying strategies for reducing the social risks. We begin by analysing the two components that make up flexicurity. Therefore, we draw on the conceptual framework of forms of flexibilization. As far as social security is concerned, we propose a set of criteria that take into account shortterm and long-term effects. We then describe the different forms of atypical work and their development since the 1980s. It is necessary to establish which individuals are particularly affected. Then we discuss the extent to which the individual forms are not just atypical but also precarious. Finally, we offer further reflections on the concept of flexicurity.


Introduction: Setting the Scene
In the past, the question of atypical work has featured in the debate on deregulation largely only in terms of employment policy aspects and the controversy surrounding the Hartz reforms. The assumption is that flexibilization will lead to the creation of more jobs (Bericht der Kommission 2002;Bundesrat 2003). In contrast, longer-term social security and employability issues have been somewhat overlooked. This imbalance in the emphasis placed on flexibility and social security is also evident in the first and second of the so-called "Hartz Laws" for "Modern Services on the Labour Market" ("Moderne Dienstleistungen am Arbeitsmarkt") that provide for an expansion in atypical forms of employment in the short term while failing to adopt measures to guard against the long-term risks implicit in this type of work.
The concept of flexicurity offers an alternative to this one-sided approach by attempting to attach equal importance to both aspects (for a summary, see Transfer 2004). In addition to taking into account the different interests of the various labour market players, a fair balance of flexibility and social security can in the long term contribute to improving labour market efficiency (OECD 2004a). Hitherto, discussion of flexicurity has focused primarily on normal employment (Normalarbeitsverhältnis), with atypical work receiving only cursory attention. This paper will attempt to remedy this conceptual oversight by identifying strategies for reducing the social risks associated with flexible forms of employment. Regulation, either in the shape of legislation or collective agreements, is one such option.
We begin by analysing the two components that make up flexicurity. In order to evaluate the importance of atypical forms of employment in companies' flexibility strategies, we draw on the conceptual framework of different forms of flexibilization developed by the OECD (1986OECD ( , 1989. As far as social security is concerned, we propose a set of criteria that take into account both short-term and long-term effects (ILO 2004).
We then move on to describe the different forms of atypical work and their development. There has been an overall increase in this type of work since the 1980s, although the different forms have not all increased at the same rate. Furthermore, it is clear that atypical work is set to become increasingly important, especially if the deregulation measures continue to be implemented. It is therefore also necessary to establish which individuals are particularly affected by the different forms. Having done this, we discuss the extent to which the individual forms can be considered to be not just atypical but also precarious. Finally, we offer some further reflections on the concept of flexicurity.

Aspects of Flexicurity
Before examining the two aspects of flexibility and social security, it is necessary to define the different types of atypical employment.

Atypical forms of employment
The literature tends to use the term 'atypical employment' as a catch-all for any form of employment that does not match the standard concept of normal employment (Mückenberger 1985;Keller/Seifert 1995;Rudolph/Schröder 1997). In this paper, we treat the term not as a normative yardstick to be understood in relation to an ideal form of work, but rather simply as a basis for our analysis, with a view to establishing the extent to which atypical forms of employment can be deemed to be precarious. For our purposes, the key aspects of 'normal work' are as follows: permanent employment contract, compulsory social insurance contributions, full-time working and corresponding pay, and a full overlap between working and employment. On the basis of these criteria, the following forms of atypical employment can be identified, occurring either separately or in various combinations: • Part-time work, • Petty (or marginal part-time) employment (geringfügige Beschäftigung) in the shape of the new 'mini-jobs' and 'midi-jobs', • Fixed-term employment (befristete Beschäftigung), • Temporary and agency work (Zeit-und Leiharbeit) including work for the recently created Personnel Service Agencies (Personalserviceagenturen -PSAs), • New forms of self-employment, including the 'Me, Inc.' (Ich-AG) programme. While the criteria listed above relate to dependent employment, the once clear boundaries between this and self-employment are becoming increasingly blurred. Consequently, we have decided to include in our analysis the new 'Me, Inc.' and 'Family Company' (Familien-AG) forms of self-employment.

Types of flexibility
The following section assumes that in an economy that is undergoing structural change and in which insecurity is prevalent, businesses need a certain degree of adaptability and flexibility. The extent and nature of this flexibility will depend on the specific circumstances of each business. Atypical forms of employment form part of the range of human resources measures that a company can deploy to increase its flexibility (Martin/Nienhüser 2002). The OECD's typology for distinguishing between different forms of flexibilization (OECD 1986(OECD , 1989) offers a sound basis for evaluating the importance of atypical forms of employment in companies' flexibilization strategies. Because it was not developed for atypical forms of employment, we will offer our own additional classification.
One key distinction that is now widely recognised is between internal flexibility and external flexibility. Internal flexibility (within a company) relies mainly on changes in working time (overtime, short-time working, working time accounts), leaving headcount largely untouched. External flexibility, on the other hand, uses the traditional approach of varying the number of employees in accordance with the company's needs (hiring and firing), and increasingly also resorting to fixed-term contracts and temporary agency workers.
Internal flexibility can be subdivided into the following categories: • Internal numerical flexibility tackles fluctuations in capacity utilisation principally by varying the number of hours worked, using working time accounts or reductions in working time introduced in order to secure jobs. • Internal functional flexibility tackles changing output requirements mainly by reorganising work processes, relying on a correspondingly multi-skilled workforce. This option is limited where strict work rules are prevalent and multi-skilling is less common, such as in the Anglo-Saxon countries, for example. 1 • One further form of internal flexibility (OECD 1989) is wage flexibility. This form is becoming increasingly popular as a result of "pacts for jobs" ('Bündnisse für Arbeit') and opening clauses (Öffnungsklauseln) allowing companies to diverge from collective agreements (Massa-Wirth/Seifert 2004). It involves measures such as the abolition of higher rates of pay for overtime or special payments, and temporary reductions in hourly pay rates. Performance-related and resultsoriented pay are also becoming more common (Bispinck/Schulten 2003). Part-time working and petty employment do not fit neatly into this classification, since both contain various elements of the different forms of internal flexibility. Consequently, we propose (internal) temporal flexibility as an additional category. This allows companies to be very precise in the way that they vary the number of hours worked in order to cover the higher workload that occurs for a limited time only during peak periods over the course of a day, week or year. Petty employment in particu-lar allows for very short working hours (e.g. two hours a day or one day a week) and this can benefit both temporal and wage flexibility. 2 External flexibility can also be subdivided into three categories: • External numerical flexibility is based on varying the headcount by hiring and firing or by using temporary agency workers and fixed-term contracts. • External functional flexibility involves improving the ability of the workforce to adapt to the external labour market. This is necessary in order to avoid mismatch problems during periods of structural change, i.e. providing employees with job skills suited to the external labour market, e.g. via the 'Transfergesellschaft', a training company that takes on workers who have been laid off and provides them with labour-market oriented training for a limited period of time. • External wage flexibility involves varying labour costs and can be influenced by wage cost subsidies. The extent to which atypical forms of employment are used depends on the relative costs of each individual form. In some cases, these forms are used as substitutes for each other and in other cases they are used to complement each other. This applies not only to the different forms of atypical employment (e.g. the relationship between part-time work/petty employment and agency work/short-term contracts) but also to the relationship between atypical work and forms of internal numerical flexibility geared towards employees on normal contracts. If we begin by comparing the different forms of atypical employment against each other, it emerges that some forms are preferred to others. Current research indicates that companies view the use of temporary agency workers as a secondary strategy to be used mainly in order to cover unforeseen short-term workload fluctuations (Bellmann 2004: 137). The use of temporary agency workers instead of fixed-term contracts cuts recruitment costs (e.g. the costs associated with advertising the post, selecting the candidates and taking them on as employees of the company) and poten-tially also reduces costs of in-firm training. On the other hand, the additional cost of hiring the workers from the agency needs to be taken into account. Both options enable companies to avoid having to pay layoff costs, and also make it possible to reduce permanent staff numbers.

Summary
Moving on to consider the relationship between internal numerical flexibility geared towards employees with normal contracts and atypical forms of employment, the first of these options appears to offer a number of advantages. In recent years, various new opportunities for internal flexibility have emerged on the German labour market as a result of collectively agreed regulations aimed at safeguarding jobs and more flexible working time arrangements (including working time accounts). The advantages vis-à-vis external forms of flexibility are obvious: there are no recruitment or subsequent layoff costs, and the company suffers no loss in its human capital and team productivity. Furthermore, these options have become cheaper for companies in recent years owing to the fact that many have done away with overtime premium (Überstundenzuschläge), for example. In addition to this, the German dual vocational training system means that employees often have a broad range of skills, making them able to perform different tasks within the company, thus ensuring internal functional flexibility.
All of the above explains why external numerical flexibility (in the form of temporary agency workers and fixed-term employment) is on average less prevalent in Germany than in other countries (Storrie 2002; European Commission 2003).

Social security versus precariousness
So far we have only discussed flexibility from the companies' perspective. As far as employees are concerned, there is a price to pay for the widely-favoured option of internal flexibility. While it does offer job security, this is achieved at the cost of financial concessions and to some extent also less favourable working hours. External flexibility measures, meanwhile, tend mostly to affect non-core workers (Randbelegschaften), with core workers (Stammbelegschaften) remaining protected against market fluctuations.
When it comes to the future of atypical forms of employment, it is necessary to establish the extent to which the characteristics of the individual forms deviate from those of the normal contracts that we are using as our yardstick. Our approach takes the long-term perspective into account and, in addition to the usual consideration of employees' situation during their working lives, also considers their situation once they have retired. Thus, it adds elements of long-term social policy to the largely short-term perspective of labour market policy.
The following main criteria are used to evaluate the extent of social security: • Income from employment should be sufficient to enable independent subsistence 3 , and social insurance contributions should ensure entitlement to a pension and to the relevant benefits in case of unemployment.
• In an economy characterised by permanent structural change and changing job and skills requirements, employability (European Commission 2001; OECD 2004a) is a key enabler of sustained employment and income. In contrast to the kind of job security that used to be implicitly considered as the norm, the emphasis on employability means that job security is no longer dependent on an individual working for a single employer. Similarly, employability plays a key role for workers moving from dependent employment to self-employment and vice versa. Consequently, further training (Weiterbildung), preferably throughout the entire course of a person's working life, is a key requirement for maintaining and improving employability. • Entitlement to social security benefits, i.e. unemployment benefit, health insurance and pensions, is acquired through the payment of compulsory social insurance contributions whilst in employment. Whereas unemployment benefit and pensions are based on the principle of equivalence (Äquivalenzprinzip) whereby benefits are theoretically calculated according to the level of contributions a person has paid into the system, health insurance entitlement is the same for everyone, irrespective of contributions paid, in line with the principle of solidarity (Solidaritätsprinzip). In the case of all the various forms of atypical employment, we need to ask whether and to what extent the different risks are covered. It is also important to note that we are referring to entitlements resulting from a person's own work and not from the employment of third parties. These criteria are partially interdependent. Thus, employability is a key requirement for income security both during and after a person's working life, while a person's income level determines the likelihood of them investing in their own human capital and hence the chances of them maintaining and improving their employability.
Atypical forms of employment are not automatically precarious, as is sometimes assumed in the public debate. Nevertheless, they do involve varying degrees of precariousness when assessed on the basis of the above-mentioned criteria (income security, employability, entitlement to social security benefits). 4 The degree of precariousness is dependent on the form and duration of the atypical employment. In general terms, it can be said that internal flexibility, while not without its risks particularly with regard to income, is relatively more secure than external flexibility (above all in the form of agency work and fixed-term employment). In the context of the flexicurity debate, it is thus necessary to identify the precarious elements of atypical forms of employment in order to make proposals as to how this precariousness can be reduced. 4 The definition of precariousness used here relates specifically to the concept of flexicurity and is thus somewhat narrower than that of, for example, Dörre et al. (2004), which includes some work content aspects.

Atypical forms of employment 3.1 Part-time work Prevalence and trends
Part-time work is a form of internal flexibilization with elements of numerical and temporal flexibility. It has increased steadily throughout all the economic cycles of the past decades and is now far and away the most important form of atypical employment. Between 1991 5 and 2004, the percentage of part-time workers rose from 14.0% to 22.8% while the overall number of people in employment experienced a slight fall (Table 1). Over this period there was little change in the highly asymmetrical gender distribution: 86% of part-time workers are women. 6 Indeed, part-time work is gradually coming to be viewed as the normal working time arrangement for women.

Precariousness
On the basis of the criteria outlined above, part-time work differs from normal work in the following respects: In view of the average number of hours worked per week (18.1 hours for women and 15.5 hours for men) (European Commission 2003: 159), income security can be problematic in certain specific circumstances: • if there are no other income earners in the household, • if a person is in part-time work for long periods, • if the part-time work is also in a low-wage sector. A pointer to the presence of income security problems is the fact that just under 15% of part-time workers say they are only in part-time work because they have been unable to find full-time jobs (Statistisches Bundesamt 2004: Table 3.17).
Part-time employment can have either a positive or a negative effect on employability. Its impact is positive when it acts as a bridge to entering or re-entering the labour market or when it allows employees with family commitments to continue working or at least offers them an alternative to taking an enforced career break. Current research indicates that while part-time employment does often fulfil this kind of intermediary function, it rarely acts as a bridge to full-time employment (Bothfeld/O'Reilly 2000). The downside to part-time work is that part-time workers have less access to further training (Unabhängige Expertenkommission 2004) and limited career advancement prospects (OECD 1999: 136). 7 This is, in fact, in contravention of the statutory requirement for part-time workers to receive equal treatment to employees on normal contracts as stipulated by the Act on Part-time Work and Fixed-Term Employment (Teilzeit-und Befristungsgestz) that came into force in 2001. It is true that the principle of equivalence does guarantee 5 This was the first year that figures were available for the whole of Germany. 6 The percentage of men in part-time work rose from 2.1% to 6.2%, while for women the figure went from 30.2% to 42.1%. 7 This is a finding of bivariate analyses which nevertheless remains sound when a third variable (such as qualifications or age) is taken into account (OECD 1999: 162). broad entitlement to unemployment benefit and pensions. However, the longer persons are in part-time employment and the lower their income, the greater the danger that the benefits they are entitled to will not be enough to ensure their subsistence. The solidarity principle, on the other hand, does ensure full entitlement to health insurance.

Petty employment Prevalence and trends
Petty employment constitutes a form of internal flexibility with some aspects of both temporal and wage flexibility. For many years, this form of atypical employment differed from part-time work with regard to income levels and working time. Since the 2003 Hartz Laws, however, only the first of these criteria has remained valid. Because of their separate trends and in particular because of differences in their importance with regard to labour market and social policy, it is necessary to distinguish between two groups: employees who have such a job in addition to a normal job (geringfügige Nebenbeschäftigung), and employees for whom petty employment is the only form of employment.
In contrast to other forms of atypical employment, until 1999, there was only sketchy information on petty employment trends owing to gaps in the time series available and inconsistencies in the methodologies used . 8 Older studies indicate that between 1987 and 1992, there was an increase of just under 36% in the number of employees exempt from social insurance contributions and persons with a second petty job (Friedrich 1995). Thereafter, this trend continued (ISG 1997) until it was temporarily reversed when lump-sum compulsory pensions insurance contributions (pauschale Rentenversicherungspflicht) were introduced in April 1999. This step resulted in changes to the composition of the main groups. It led to a significant fall in the number of employees who had a subsidiary job in addition to a normal job. This fall was partly counteracted by a rise in the number of people for whom petty employment was the only source of employment, meaning that the overall change was relatively modest (Schwarze/Heineck 2001).
The second Hartz Law for Modern Services on the Labour Market (Hartz II) came into force on April 1, 2003 and marked an institutional and conceptual turning point. It replaced the old forms of petty employment (earning up to 325 euros a month) with mini-jobs (up to 400 euros a month) and midi-jobs (401 -800 euros). 9 These reforms represented a change of course for the government. While the 1999 8 Since 1999, the Federal Employment Agency (Bundesagentur für Arbeit) has compiled statistics on petty employment based on the number of employers registering such workers with the social security system. 9 Mini-jobs are tax-free for employees, while employers have to pay 23% of the employee's wage as pensions (12%) and health insurance (11%) contributions and an additional 2% as wages tax. As for midi-jobs, employees' social insurance contributions are staggered, reaching the full rate of 21% for people earning 800 euros. Employers pay their usual percentage of the contributions. The fact that expenses for house maids (Haushaltshilfen) are partly tax-deductible (up to 10% of the cost, up to a maximum of 510 euros per year) is designed to encourage people who have not been declaring this kind of income to do so. regulations sought to reduce the number of people in petty employment, the new legislation aims to increase it.
The available data have improved significantly since the introduction of the Hartz Laws, since as the body responsible for pensions insurance, a special Federal Pension Fund (Bundesknappschaft), now records the number of cases. 10 Since the introduction of mini-jobs, there has been a significant rise in the number of people (cf. Table 2). 11 This trend can largely be put down to people who have another job in addition to their normal activity and do not pay social insurance contributions for their second job, 12 including a significant number of school and university students and pensioners. 13 The age distribution of this group is relevant, with large numbers being concentrated at the lower end (under-25s: 18.2%) and upper end (over-60s: 19%) of the scale. Finally, the majority of workers in this form of employment are women (64.2%). The official statistics do not distinguish between these groups. A sample survey indicated that pensioners and school and university students make up at least a quarter of the employees in this group (Fertig et al. 2004).
In contrast, it is much harder to come by adequate data for midi-jobs. Preliminary data for the end of 2003 put the number of people in this form of employment at 669,000, and in the vast majority of cases their income only occasionally falls into the 401 -800 euro bracket (Statistisches Bundesamt 2004).

Precariousness
This section will focus on people for whom mini-jobs are the only source of employment, since people for whom they are just a subsidiary job are already covered by the social security system through their main job; school and university students and pensioners have their own corresponding social security protection. So what of the people for whom mini-jobs are the only source of employment and who do not acquire social security entitlements through another job or have any other entitlement to social security protection (for example the entitlements enjoyed by school and university students and pensioners)? They form part of the labour market, but do not have the same statutory social security entitlements as employees in normal jobs. Owing to the low level of the contributions that they pay, their pensions are insufficient to ensure their subsistence 14 and they do not earn an individual entitlement to health insurance. Furthermore, only people earning 401 euros or more, i.e. employees with midi-jobs, are entitled to claim unemployment benefit. Consequently, it can be said that this type of work is not sufficient to ensure independent subsistence unless a person has more than one mini-job on the go at the same time. In addition, these people have limited employability, since they hardly receive any further training at all (Unabhängige Expertenkommission 2004: 120).

Fixed-term employment Prevalence and trends
Fixed-term jobs constitute a 'classical' form of external numerical flexibility that is used to cover situations when a temporary need for additional workers arises. Unlike normal jobs, fixed-term jobs are terminated at an agreed time and without the need to follow legal procedures. Between 1991 and 2003, the percentage of people in this form of employment rose only slightly from 6.4% to 7.8%. 15 This is a rather surprising figure, owing to the fact that since the Employment Promotion Act (Beschäftigungsförderungsgesetz) came into force in the mid-1980s, the statutory regulations govern-14 This is still true even when employees take advantage of the option of increasing their reduced pension contributions to the full rate. These voluntary contributions actually happen in less than 10% of cases (Fertig et al. 2004). 15 These figures are based on calculations kindly made available to us by Rudolph (2004). They only cover private sector blue-collar and white-collar workers, and do not include trainees and public sector employees. The higher percentages recorded by other sources (Europäische Kommission 2003: 118) can be attributed to the inclusion of trainees.
ing fixed-term employment without any objective reasons have been repeatedly relaxed. 16 As far as gender distribution is concerned, there are few differences between the sexes (women 8.2% and men 7.5% in 2003). However, age distribution is another matter, with this form of employment being clearly more common among younger age groups (under-30s). This distribution pattern indicates that an increasing number of people's first experience of regular work is on a fixed-term contract (Erlinghagen/Knuth 2002;. The relaxation of the conditions governing fixed-term contracts has not only increased HR managers' scope for using flexible forms of employment but has also provided them with an additional instrument of staff probation and selection.

Precariousness
As far as social security entitlement is concerned, fixed-term employment in itself need not necessarily give rise to any fundamental problems compared with normal employment. The problems only arise when there are gaps between several periods of fixed-term employment, leading to an increased risk of income insecurity both during a person's working life and, as a result of their lower pension contributions, once they have retired. These problems do not occur when fixed-term contracts act as a bridge leading to permanent jobs as is often the case during the transition from the education and training system to the world of work. After falling in the second half of the 1990s, the percentage of people in this specific group who go on to get a permanent job has more or less stabilised at around 46% (Bellmann/Alda 2004: 258).
However, the overall percentage of people who go on to get a permanent job is very different, standing at approximately 25% in western Germany and about 17% in eastern Germany according to recent estimates (Bellmann/Alda 2004: 258). Another study puts the figure at nearer 40% (Boockmann/Hagen 2004). Furthermore, unemployment is relatively more common than among people with normal jobs, and the average time taken to find work is also significantly longer (Bielenski et al. 2003). This can cause problems as a result of the fact that the Hartz Laws have reduced the period of time during which unemployment benefit may be claimed. All this makes it clear that there is a fine line between fixed-term employment acting as a 'bridge' towards the primary labour market and the danger of people having a 'succession of fixed-term contracts' thereby damaging their prospects of finding permanent employment (Giesecke/Groß 2002).
There is a shortage of empirical data on the extent to which employees on fixedterm contracts have access to further training in order to improve their employability (one of the few exceptions is OECD 1999). Human capital theory calculations suggest that businesses invest less in fixed-term workers owing to the shorter amortisation period, despite the fact that the Act on Part-time Work and Fixed-Term Employment stipulates that they should receive the same treatment as permanent employees. The 16 e.g. the extension of the maximum duration of a fixed-term contract, the abolition of the ban on employing people on successive fixed-term contracts, and the introduction of unlimited fixed-term contracts for employees above the age of 52. opportunities to participate in publicly funded further training courses during periods of unemployment between fixed-term contracts have decreased, since measures of this kind are no longer being so heavily promoted (Dobischat 2004). The suggestion that employees could pay for their own further training runs into difficulties if we take into account the unsteady nature of their earned income and the associated relatively low average income level.

Temporary agency work Prevalence and trends
Like fixed-term employment, temporary agency work, including work for Personnel Service Agencies, is a form of external numerical flexibility. One of its specific characteristics is that there is a kind of triangular relationship between the employee, the temporary employment agency that is their official employer, and the user company, that has the right to instruct them as to the work to be undertaken.
Temporary agency work increased steadily from the time when it was first authorised by the 1972 Loan Worker Employment Act (Arbeitnehmerüberlassungsgesetz) until the cyclical downturn in 2001. While in 1991 it still accounted for less than 0.5% of all work, by 2003 the figure had passed the 1% mark. The fact that temporary agency work fell more than regular employment between 2002 and 2003 indicates that it is very sensitive to cyclical fluctuations compared to normal work, an assertion that is backed up by multivariate analyses (Bellmann 2004: 140). The considerable publicity that this form of employment has attracted can be put down to its sustained high growth rates over the last three decades rather than the absolute number of temporary agency workers, as well as to the high percentages of agency workers in some other countries (in particular the Netherlands) and the expectations concerning additional employment opportunities generated by these figures (for a summary of trends in this area, see Schröder 1997).
The successive extensions of the maximum period for which a temporary agency worker may be hired out to a company 17 did not result in a corresponding increase in the number of longer-term work placements, which in fact experienced only a moderate rise. In 2002, just under 40% of temporary agency work placements were for a duration of three months or more, evidence that short-term placements (of up to three months) continue to be the preferred option (Statistisches Bundesamt 2004: 92). The mid-1990s saw a slight rise in the percentage of women doing this type of work, but the figure is still below 25%. Furthermore, on average, temporary agency workers have fewer qualifications (Rudolph 2003: 21).
In April 2003, the Hartz Laws introduced a new form of temporary agency work known as Personnel Service Agencies (PSAs). These are designed initially to offer unemployed people temporary employment opportunities with agencies with a view to helping them enter the primary labour market in the longer term (frequently referred to as the 'glue effect' or 'Klebeeffekt'). So far at least, there are two respects in which PSAs have failed to live up to the high hopes that they originally raised. The overall 17 Since the mid-1980s the period has been extended from originally three months to six, then twelve and now 24 months for more details see Rudolph (2003: 17). number of people employed by PSAs has remained low, and, disappointingly, only just under a third of these workers have gone on to find jobs for which social security contributions are payable. 18 Precariousness Temporary agency work is highly flexible in nature, not only with regard to the work undertaken at the assigned firm but also as far as the employment with the temporary work agencies is concerned, since in practice these offer little prospect of stable employment. The significant fluctuations in this form of work are on the one hand due to its function as a 'bridge' leading to stable employment. In the past, the percentage of employees moving on to steady jobs has fluctuated between 30% and 45% (Jahn/ Rudolph 2002: 4). In addition, temporary agency work is used to cover seasonal workload peaks and offers no guarantee of continued permanent employment during periods of low demand. The initial controversy surrounding this form of work hinged on the issue of whether it is the agency's or the user company's terms and conditions of employment that should apply to temporary agency workers, particularly as far as pay is concerned. It appears that the negotiation of the first collective agreement for the sector has resolved this dispute, at least for the time being (Bispinck/WSI-Tarifarchiv 2004). Hitherto, temporary agency workers' pay has been considerably lower than that of permanent employees of the user company doing equivalent work (Kvasnicka/Werwatz 2002;Bellmann 2004). The extent to which the collective agreement is able to close the income gap in practice remains to be seen.
As far as social security is concerned, temporary agency workers have the same entitlements as normal workers as long as they are not doing part-time work, which is in fact rarely the case (Jahn/Rudolph 2002). The Hartz Laws reformed the Loan Worker Employment Act by (Arbeitnehmerüberlassungsgesetz) stipulating that tempo- 18 The fall in the number of PSA employees at the beginning of 2004 can be mainly attributed to the fact that the largest provider, Maatwerk, filed for insolvency at this time. rary agency workers must receive the same key terms and conditions of employment as employees doing equivalent work in the company to which they have been hired out.
As far as employability is concerned, empirical data support the conclusion that temporary agency workers participate in further training less frequently than employees with regular contracts (Unabhängige Expertenkommission 2004: 52). This finding is backed up by human capital theory arguments, according to which user companies can be expected to err on the side of caution when it comes to investing in further training, owing to the fact that agency workers are on average employed for short periods of time.
On the other hand, PSAs have a statutory requirement to provide their employees with further training (Par. 37c SGB III). Nevertheless, there are a number of practical difficulties associated with organising training for PSA employees in the periods between their work placements (e.g. unpredictable duration of work placements, management problems, insufficient modularization of training programmes). This problem has been demonstrated by the similarly structured "START part-time work" model project ("START Zeitarbeit") (Weinkopf/Krone 1995).

New forms of self-employment Prevalence and trends
Labour market policy has introduced several instruments designed to encourage people to move from unemployment to self-employment. In addition to the 'traditional' bridging allowance (Überbrückungsgeld) that was introduced in 1986, the Hartz Laws created a complementary business start-up grant (Existenzgründungszuschuss) that forms the basis of the 'Me, Inc.' and 'Family Company' forms of self-employment. The aim of this measure is to ease the transition from unemployment to selfemployment and to prevent people from not declaring their earnings. The start-up grant is a monthly payment that gradually decreases over time and may be claimed for up to three years as long as the claimant's annual income does not exceed 25,000 euros. 19 The number of people taking up this option has been rising rapidly and by the end of 2004 had already exceeded the number of people claiming the bridging allowance. 19 The Me, Inc. structure does not allow the self-employed person to employ anyone else, although the 'Family Company' does permit other family members to participate.
However, in itself the number of participants tells us very little about how successful the measure has been. In order to assess whether Me, Incs really provide a route out of unemployment, it is necessary to take account of how long the participants remain self-employed, as well as the drop-out rate. It is important to distinguish between the reasons people have for leaving the scheme, which include their income rising above 25,000 euros, transfer to dependent employment, or failure of their Me, Inc. business. Other factors to be taken into account when evaluating this measure, in addition to how long the participants remain self-employed (the 'survival rate'), are the potential windfall and substitution effects (Mitnahme-und Substitutionseffekte) (Koch/Wießner 2003;OECD 2004b: 61). However, the fact that the measure has only been up and running for a short time means that it is still too early to make any empirically based conclusions on these matters.

Precariousness
Valid data for survival rates and employment effects will only become available at the end of the first three-year grant period. Preliminary findings indicate, however, that a significant percentage of the people claiming the grant have already abandoned the goal of self-employment before the end of the three-year period. More than half have become unemployed again, and a third of them report that they now have debts resulting from their brief period of self-employment (Wießner 2005). While they are claiming the grant, people who set up a Me, Inc. are required to make compulsory pension contributions at a reduced rate, but once the grant period has elapsed they are responsible for making their own provisions (Schulze-Buschoff 2004). Health insurance contributions are voluntary and, once again, on favourable terms, while no unemployment insurance contributions are paid. Income security problems only really cease to be an issue if an individual's annual income (including allowance) substantially exceeds the 25,000 euro threshold. Furthermore, problems can arise if people's annual income (including income from dependent employment) fails to rise significantly after the end of the grant period (Förderzeitraum). Last but not least, it is up to the 'new' self-employed to make their own arrangements for maintaining or improving their employability.

Conclusions
The politically-motivated promotion of atypical forms of employment has given rise to a number of labour market and social security problems related to the degree of precariousness present in this type of work. Further training is increasingly important owing to structural change, and people in atypical forms of employment have less access to it than people with normal jobs. Meanwhile, this type of employment also means that people end up with inadequate pension entitlements. It is true that, as has been shown, these failings do not affect all atypical forms of employment to the same extent. However, overall it can be argued that the rise in flexible forms of employment has been accompanied by lower levels of social security protection, and in the longterm, this could also have a negative impact on the efficiency of the labour market because of a shortage of qualified labour. The dangers increase the longer a person remains in this form of employment.
The concept of flexicurity aims not to reduce flexibility but rather to diminish the precariousness associated with it. In the case of atypical forms of employment, there are two obvious strategic approaches to pursuing this goal. The first is to attempt to use forms of internal flexibility instead of external flexibility wherever possible (e.g. working time adjustments introduced to secure jobs or working time accounts instead of temporary agency workers or fixed-term contracts). The second is to use flexicurity measures to reduce the precariousness associated with atypical forms of employment (Keller/Seifert 2002): Low incomes are already being supplemented by transfer payments in order to ensure that they attain subsistence levels. Furthermore, the right to return from parttime to full-time work as introduced in the Netherlands offers another way of tackling income security problems, although this is not provided for by the German Act on Part-time Work and Fixed-Term Employment. As far as fixed-term employment is concerned, income security is only a problem in certain specific circumstances, since short periods of unemployment can be covered by unemployment benefit entitlements. Meanwhile, the recent collective agreements, if properly implemented, promise to reduce the income disparity hitherto experienced by temporary agency workers.
Access to further training is key to maintaining an individual's employability. Possible approaches to tackling further training problems could include giving all the workers in a company the right to receive training, the promotion of publicly funded training, or organising training at a supra-company level. Organising their own training is problematic or simply not an option for many people who lack the resources to do so, either because of their low incomes or also because of their unstable employment situation which may be punctuated by lengthy periods of unemployment. The restricted access to further training experienced in practice, for example, by part-time workers could be tackled by introducing opportunities of universal access, particularly through collective agreements. Offering further training to people while they are out of work in the gaps between fixed-term contracts, and reversing the current trend to reduce the amount of publicly funded training available, would help to create a workforce capable of meeting changing skills requirements. Further training for temporary agency workers would be best tackled by the creation of sectoral training funds (Branchenfonds) (Unabhängige Expertenkommission 2004: 241).
As far as social security is concerned, the forms of atypical employment enjoy different levels of protection, with temporary agency workers at the top and people in petty employment at the bottom of the scale. Solutions inherent to the system (e.g. regarding the danger of part-time workers suffering poverty during old age) could be based on flexible entitlements (flexible Anwartschaften) that would allow the acquisition of pension entitlements over a defined number of years for people in forms of employment in which they pay few contributions. This would mean that breaks in employment for whatever reason (e.g. having a family or further training) would no longer result in substantial reductions in pension entitlement . A more far-reaching option is provided by basic minimum income models as seen, for example, in Switzerland (Brombacher-Steiner 2000).
Little or no attention has been paid in the current debate to the question of which costs of atypical employment are externalised either through inadequate pension contributions (Steiner/Wrohlich 2005) or through the lack of contributions to unemployment insurance. Unless it succeeds in its ambition to create more jobs, the shortterm promotion of these forms of employment will in the long term undermine pension funds. At the same time, it will reduce the ability of an increasing number of employees to make their own provision for retirement.
The answer to the externalisation question depends principally on the additional or net employment impact of atypical work. This is hard to quantify empirically, but might be expected to be relatively minor. The substitution effects that arise as a result of the replacement of regular jobs by mini-jobs (Schupp/Birkner 2004;Steiner/ Wrohlich 2005) or temporary agency work (Promberger 2005) lead to a loss of revenue for the social security system. However a more detailed analysis of the consequences of this for the institutions of the social security system would go beyond the scope of this paper.