The quasi-market of employment services in Italy

The quasi-market of employment services in Italy

+Model JPO-6548; No. of Pages 22 ARTICLE IN PRESS Available online at www.sciencedirect.com ScienceDirect Journal of Policy Modeling xxx (2019) xxx...

987KB Sizes 0 Downloads 0 Views

+Model

JPO-6548; No. of Pages 22

ARTICLE IN PRESS Available online at www.sciencedirect.com

ScienceDirect Journal of Policy Modeling xxx (2019) xxx–xxx

The quasi-market of employment services in Italy Francesco Pastore 1 Department of Economics, Corso Gran Priorato di Malta, I-81043, Capua, Caserta, Italy Received 18 October 2018; received in revised form 12 May 2019; accepted 25 June 2019

Abstract This paper aims to study pros and cons of the first experiment of quasi-market in the provision of employment services in Italy: the Lombardy DUL (Dote Unica Lavoro). The program, which has inspired the 2015 national reform within the Jobs Act and lines towards the recent experience of several Anglo-Saxon countries, has revitalized the sector by providing important job opportunities to jobless workers. We find the typical problems of quasi-markets (lion’s share of private organizations; cherry picking; gaming and asymmetric information). However, different expedients were devised in the program to minimize these shortcomings. The empirical analysis suggest that such phenomena are at a physiological level. Analysis of the determinants of completing successfully the program provides non-trivial results as to, among others, the role of organizations of different ownership type and of the type of services provided. On these we base our policy advice. © 2019 The Society for Policy Modeling. Published by Elsevier Inc. All rights reserved. JEL classification: H44; H52; H76; I38; J68; R23 Keywords: Public employment services; Quasi-markets; Cherry-picking; Gaming; Lombardy region; Jobs Act

Abbreviations: ARIFL, Agenzia Regionale per la Istruzione, Formazione e Lavoro (Regional Agency for Education, Training and Employment); COB, comunicazioni obbligatorie (En.Tr.: compulsory registrations to employment offices); DUL, Dote Unica Lavoro (En.Tr.: single working voucher); EYG, European Youth Guarantee; NEETs, Not in Employment Education or Training; PES, Public Employment Services; PIP, Piano d’Intervento Personalizzato (En. Tr.: personalized action plan); VET, Vocational and Training System. E-mail address: [email protected] 1 www.iza.org/profile?key=692. https://doi.org/10.1016/j.jpolmod.2019.06.008 0161-8938/© 2019 The Society for Policy Modeling. Published by Elsevier Inc. All rights reserved.

Please cite this article in press as: Pastore, F. The quasi-market of employment services in Italy. Journal of Policy Modeling (2019), https://doi.org/10.1016/j.jpolmod.2019.06.008

+Model

JPO-6548; No. of Pages 22 2

ARTICLE IN PRESS F. Pastore / Journal of Policy Modeling xxx (2019) xxx–xxx

1. Introduction The aim of this essay is to assess the effectiveness of the DUL (Dote Unica Lavoro), a program implemented in Lombardy region, under the aegis of the European Social Fund, to introduce a quasi-market organization of employment services. The program was devised to overcome the difficulties of staffed employment services in a local labor market which is one of the most dynamic in the country. A number of previous interventions unsuccessfully aimed to increase the effectiveness of employment services nationwide. First, it was the end of state monopoly by the Treu Law of 1997. The Biagi Law of 2003 stated the full equality of rights of private (forprofit and non profit) and public employment services (PES). Despite a certain tendency of the local population in favour of private agencies, still only a small number of NEETs (Not in Employment Education or Training) were using placement services and training supplied in the system before the reform. The intuition of the DUL to stimulate the use of employment services by the possible customers was to assign vouchers (the dote is a voucher) of different amount to jobless workers according to their actual need, as assessed by a profiling based on objective criteria: duration of the unemployment spell, education level, age, gender. The voucher is a tool for users to choose the best services available on the market. In addition to vouchers, accredited organizations may claim a special reward for every DUL completed “successfully”, namely that leads to employment for at least six months. The analysis is essentially descriptive, due to data limitations, and aims to study for the first time by means of administrative data some characteristics of the program as well as the determinants of success in completing successfully the program.2 We provide important insights as to the effectiveness of the program in reaching its aims and at the same time in removing some of the most important shortcomings that economic theory has brought to the fore on quasi-markets (Le Grand, 1991). In fact, different tools were devised in the programme to address the typical shortcomings of quasi-markets. With the end of the traditional Keynesian approach to economic policy, the goal of full employment has been gradually replaced by that of employability (Centeno & Stewart, 2013). Expansionary fiscal and monetary policy have been slowly replaced not only by passive income support schemes, but, even more so, by active guidance, vocational training and self-employment policies. From a principally macroeconomic perspective to a microeconomic one. At the center of these policies there is not only vocational training, but also the education system: the final goal of employability can be achieved when an efficient system of education and vocational training is created (VET system); hence the ensuing need to improve public education (in particular, high technical and vocational education), but also training and, importantly, from the point of view of this study, employment services. Recently, university education has become a key factor of success in the school to work transition. Nonetheless, education attainment itself, which is in Italy one of the OECD lowest, is not everything. In fact, in countries where education follows a sequential, rather than a dual principle,3 the mission of the education system is simply general education, leaving to post-school and post-university steps the task of promoting general and job specific work experience, without 2 Montaletti (2015) provides a causal analysis of the impact of the policy as based on quasi-experimental methods of randomized comparison between a target and a control group. Here this is not allowed due to the lack of suitable data. 3 Though, the Good School reform of 2015 has eventually introduced a short period of compulsory work related learning for high secondary school students.

Please cite this article in press as: Pastore, F. The quasi-market of employment services in Italy. Journal of Policy Modeling (2019), https://doi.org/10.1016/j.jpolmod.2019.06.008

+Model

JPO-6548; No. of Pages 22

ARTICLE IN PRESS F. Pastore / Journal of Policy Modeling xxx (2019) xxx–xxx

3

which human capital is incomplete and employability is compromised. In dual education systems (e.g. Germany), classroom education and vocational on-the-job training happen at the same time: in this case, the education system’s mission is forming all-round human capital (Pastore, 2015a, 2019). In sequential systems, the task of generating work-related competences is left to the labor market or the VET system (depending, among others, on the efficiency of Job centres). In Italy, sequential education – characterized by a low integration with enterprises also for high technical and vocational schools – is accompanied by a labor market that is slowly becoming flexible and by an unsatisfactory VET system. Within this context, efficient employment services are a key issue. Instead, only about 3.5% of new hires happen through public and private employment services, which compares to about 7% of the UK and 13.5% of Germany (Giubileo, 2012; Pastore, 2013). In addition, at least from the 2001 reform of Title V of the Constitution, which attributed the jurisdiction for the supply of post-school vocational training to Regional authorities, there has been a “balkanization” of employment services. Some Regions (e.g.: Lombardy, Veneto and Emilia Romagna) showed a better organizational and innovation capacity than others (especially the Southern Regions) (Giubileo, Leonardi, & Pastore, 2013). Besides, traditionally, Northern regions are able to supply more jobs than Southern regions (Mauro, 2004). Our conclusion from the policy point of view will be to show that the DUL policy is an important step forward to address many of the shortcomings of the Italian system of employment services, although at some cost, such as an increased amount of public spending of the overall service and the lion’s share of the private sector. Nonetheless, several corrections could be devised to further control these shortcomings. This essay is structured as follows. Section 1 discusses the aims of an evaluation of a program like DUL, with all its shortcomings and merits. Section 2 describes the way of working of the program. Section 3 proposes a descriptive analysis of several aspects of the program. Section 4 focuses in part one on the methodology adopted for the econometric analysis of the determinants of the probability to successfully complete the program. The description of the data set is provided in Supplementary Annex 1. Another subsection presents the main results. Section 5 supplies our policy discussion of the results. Some concluding remarks follow. 2. The aims of the assessment 2.1. Shortcomings and merits of the policy DUL is a complex pro-active program, which implies, first of all, a transformation of the market structure and re-organization of employment services. By introducing a “quasi-market”, the program aims to “reactivate”, so to say, not only the NEETs, but also the supply itself, following a wave in a growing number of countries, starting from Australia, UK, Netherlands, Germany. DUL gives to beneficiaries the possibility to freely choose the accredited operator they prefer. One of the main limitations of the public sector monopoly is that users cannot choose and, usually, this tends not only to limit the effectiveness, but also the quality of the public service4 . The quasimarket structure allows otherwise nonpaying demand to express their preferences by empowering 4 Arrow (1963) is among the first to outline the shortcomings of the production of public services when users cannot choose. His analysis refers to the health sector, but can be easily extended to all public services managed similarly. Le Grand (1991) and Bartlett and Le Grand (1993) are among the first to show the way of working - advantages and limitations - of quasi-markets in the implementation of social policies in Great Britain in the Eighties.

Please cite this article in press as: Pastore, F. The quasi-market of employment services in Italy. Journal of Policy Modeling (2019), https://doi.org/10.1016/j.jpolmod.2019.06.008

+Model

JPO-6548; No. of Pages 22 4

ARTICLE IN PRESS F. Pastore / Journal of Policy Modeling xxx (2019) xxx–xxx

it with a voucher to “buy” services from accredited operators. Following the Biagi Law of 2003, not only public, but also private (forprofit and nonprofit) operators enter the competition to supply the services requested. In this market structure, when a “users sovereignty” is (almost) restored, it is clear that – moved by a quasi-invisible hand – the same accredited operators will develop all the tools – the human and professional skills – necessary to produce the best employment services they can to satisfy users and, at the same time, gain the voucher. Quasi-markets enhance competition among accredited operators to gain the maximum number of users they can. The idea to create a quasi-market was also at the heart of Decree n. 150/2015 approved by Renzi’s Government within the Jobs Act5 for the redefinition of the institutional and regulatory framework in the field at a national level. Undoubtedly, DUL inspired the national reform of the sector, although the national level differs in some aspects from the Lombard DUL, which has generated some debate among experts and policy makers. The main difference between DUL and national setting is in the role of the public versus private sector. The idea behind the national reform is that the state sector should maintain a supervisory role of monitoring, which implies the task of issuing the assegno di ricollocazione, name given to vouchers in the national system, instead of dote. In the national framework, it is the state who should profile users in groups of need based on objective indicators, although like dote, also assegno can be spent with any public and private accredited operator. This should reduce the conflict of interest that might arise if private companies would profile users themselves. However, since the state sector is often inefficient, job seekers may ask also private accredited operators to be profiled, like in the DUL, after two months have uselessly passed from the first request to a PES (see, for instance, the re-joinder by Leonardi, 2015, one of the writers of the law, to Bocchieri, 2015, the director of the program in the Lombardy region). Assessing the DUL effectiveness is interesting not only from a regional viewpoint, but also from that of central government and other advanced countries interested in similar reforms to make PES work. On one hand, quasi-markets increase the so-called x-efficiency of operators and the market, on the other hand, however, there are some shortcomings, well-known since their first introduction in the supply of public services in Great Britain in the Eighties. However, these shortcomings have not prevented quasi-markets from spreading in all advanced economies thereafter, probably for a lack of suitable better alternatives. The first shortcoming is that efficiency will increase under the condition of neither imperfect nor asymmetric information, which could prevent users from freely choosing the best operator. Moreover, while, on the one hand, state monopoly prevents the entrance of other operators, therefore constraining competition, on the other hand, an excessive fragmentation of supply could prevent the exploitation of economies to scale. Giubileo and Pastore (2013a, 2013b) and Giubileo et al. (2013) outline the most common and important risks in a quasi-market context: cherry picking and gaming. Table 1 presents a snapshot of the main shortcomings, together with the solutions adopted, also within the DUL. The empirical analysis will provide evidence of both. Especially private employment agencies can activate mechanisms of cherry picking or creaming of the easiest to place beneficiaries. This happens sometimes through the so-called “refusal” of the

5 The Jobs Act included a number of reforms of the labor market in continuation with the structural reforms of Monti’s government and especially the Fornero Law of the labor market. For an assessment of the impact of the 2011-2013 reforms, see Annicchiarico, Di Dio, and Felici (2013).

Please cite this article in press as: Pastore, F. The quasi-market of employment services in Italy. Journal of Policy Modeling (2019), https://doi.org/10.1016/j.jpolmod.2019.06.008

+Model

JPO-6548; No. of Pages 22

ARTICLE IN PRESS F. Pastore / Journal of Policy Modeling xxx (2019) xxx–xxx

5

Table 1 Shortcomings and advantages of DUL. Shortcomings

Solutions

A) Loss of economies to scale due to increasing fragmentation of supply; B) Lion’s share of the market to the private sector

A) Increase in the productivity of operators and the quality of services to compensate for higher costs; B) In the long run, also the public sector will improve its skills and gain market shares;

C) Opportunistic behavior, especially of the private sector: CA) cherry picking or creaming of the easiest-to place users and parking of the least easy to place

CB) Gaming with users in favor of rent seeking behavior;

D) Asymmetric information about the quality of the services provided might prevent users from choosing the best supplier.

E) Private employment agencies might have a conflict of interest in profiling

CA1) Users are divided in groups according to needs; CA2) and receive vouchers of increasing amount according to the need; CA3) rating of operators is made by group of needs (internal rating); CB) Bonuses are supplied only to organizations which are able to place workers for a long period of time (6 months–1 year) so to have a cost higher than the benefit of gaming; D1) accreditation of the operators which reach a given standard; D2) continuous monitoring of all service providers; D3) internal rating of operators based on evaluation of their performance with users belonging to different groups of need; D4) rating by an independent third agency. E1) profiling is made by PES with a role of monitoring and evaluation of the services provided; E2) profiling on objective, transparent and verifiable criteria.

hardest to place individuals, pushed towards public or non-profit organizations according to the case (elderly, illiterates, without compulsory education, very long-term unemployed, disabled and so on). This helps maximizing the gain per effort made in placement, job guidance and vocational training, because: 1 younger beneficiaries tend to accept temporary jobs more easily; 2 those with higher education find more easily a good match; 3 private organizations aim to gain positive results in the shortest possible time. Of course, a certain degree of creaming is physiological, because the best qualified beneficiaries might themselves prefer private operators, as they are considered more effective and supply services of higher quality than public or nonprofit operators. Therefore, it is not easy to understand whether creaming is actually the result of an incorrect behavior by the private operator, to be censured. The program tries to discourage pathological creaming with different tools, by: • assigning a higher voucher to the more in-need-groups; • assessing placement rates by in-need-group, which allows a fairer rating of operators, taking into account their ability to place also the hardest-to-place beneficiaries. Please cite this article in press as: Pastore, F. The quasi-market of employment services in Italy. Journal of Policy Modeling (2019), https://doi.org/10.1016/j.jpolmod.2019.06.008

+Model

JPO-6548; No. of Pages 22 6

ARTICLE IN PRESS F. Pastore / Journal of Policy Modeling xxx (2019) xxx–xxx

Gaming is the tendency of (especially private) operators to agree with the beneficiary a deal to get the reward for successful completion of the program even if the terms for its payment are unmet: e.g. to simulate a labor contract that is discontinued soon after obtaining the benefit. A series of tools are devised to prevent gaming. First, the incentives recognized for the labor contracts signed, which are additional to the voucher for the supply of placement services, are awarded only for contracts of at least 6-months. This makes gaming not-convenient, since the cost of the contract would outweigh the benefit. Evaluation is important to make quasi-markets work efficiently based on the principles of New Public Management. First, policy makers (as well as median voters) may be interested to understand if the program has had any impact on employment and at which cost. In particular, it matters whether the new jobs are additional or substitute to those that would have been created anyway. Second, the efficient working and continuous improvement of the policy would need detailed and sound information so as to reduce the asymmetric information between operators and beneficiaries about the quality of services. This is a necessary condition for beneficiaries to choose the best among accredited operators not as based on alleged subjective assessment, but on objective criteria and transparent, sound, evidence-based analysis. In a sense, evaluation should be conceived as an integral part of the program, together with the accreditation and rating of operators. Accreditation guarantees a minimum quality standard, but it neither allows to verify the best/worst operators nor to inform the choices of users. A rating system of operators would be important. However, to ensure that rating is objective, verifiable and up-to-date, it must be based on rigorous assessment studies, carried out according to protocols agreed between policy makers and operators. Only in this way beneficiaries will be able to make conscious decisions, a necessary condition to increase the allocation efficiency. 2.2. Aims of monitoring and evaluation Quasi-markets based on the above principles allow reaching several desirable outcomes of employment policies, such as increasing: a) b) c) d)

NEETS’ skills; employability; especially in a permanent job; the efficiency of all accredited operators.

DUL seems to possess all the necessary characteristics to effect each of these outcomes. Montaletti’s (2015) study provides a counterfactual pattern with DUL users considered as “target group” and a “control group” formed by unemployed people, selected from the Compulsory Communication (comunicazioni obbligatorie) database among those who lost their job in the same period in which the policy was implemented. Moreover, the control group was selected according to employment and demographic characteristics similar to those of the target group. The author, showed that DUL beneficiaries had systematically higher chances to find a job than others, who did not participate in the program. a) Workers skills In the German tradition, the first purpose (often forgotten) of pro-active schemes and, therefore, also DUL, would not be employment itself, which is rather the final outcome of the policy, but work related skills which are a pre-condition for increasing employability. Please cite this article in press as: Pastore, F. The quasi-market of employment services in Italy. Journal of Policy Modeling (2019), https://doi.org/10.1016/j.jpolmod.2019.06.008

+Model

JPO-6548; No. of Pages 22

ARTICLE IN PRESS F. Pastore / Journal of Policy Modeling xxx (2019) xxx–xxx

7

However, one problem with this approach is that it is hard to measure skills. Often there are no ex ante and ex post measures, that are basic elements for a proper policy assessment. In the future it will be necessary to collect this kind of information, although first of all it is necessary to exactly clarify what to measure. It could be job productivity or also something else. In the German case, skills are often certified by professional qualifications, an objective indicator. Alternatively, one way to measure skills is to verify whether they have increased job opportunities. But this means employment. That is why, in the Anglo-Saxon tradition, the skills of users are not important. Evaluators follow a black box approach: it is not interesting “how” the accredited operator creates jobs, whether it delivers vocational training or not and whether skills increase or not. The operator must find employment to beneficiaries: this is the only thing that matters when paying a bonus. b) & c) Effects on employment The key goal of employment policies and evaluation studies is employment. All the observers and, in particular, policy makers are interested to know: How many jobs have been created with this program? Are they additional or substitutive jobs to those that the market would still have created by itself? Are they fixed-term or permanent jobs? Important decisions may depend on that, such as whether or not to continue the program, for how long, whether in its current or in a modified shape, with which resources, more, less or equal to those already existing, and so on. Montaletti’s (2015) microeconomic assessment applies a randomized study method. The microeconomic assessment may also be subjected to the confirmation of macroeconomic analysis. If the jobs created are really additional to the existing ones, a rise in the overall share of employment should be observed. The difference-in-difference approach is the most common: the employment rate is compared before and after the program has taken place. Then, the investigator may compare whether a similar increase occurred also in other areas of the country, especially contiguous regions, not covered by the program. This gives a measure of the DUL impact compared with other concomitant or disturbing factors. d) Impact on the efficiency of employment services As already noted, one of the most important aspects of DUL is to transform the sector of labor intermediation from passive into dynamic, where everyone (not only private operators, but also the public ones, in a longer run) have a convenience to supply services to the public. In this sense, a shift from a bureaucratic management style to a management users and market oriented one should be observed. The pre-reform Italian regulatory framework concerning labor intermediation had features that made it production oriented: a) rigid attitudes by all operators; b) bureaucratic rules that make it unattractive to produce services for public and private operators; c) little convenience by users, both from the demand (companies) and the supply side (unemployed, NEETs); d) insufficient freedom to choose among alternative operators for beneficiaries. Yet, as Weishaupt (2011), Larsen and Vesan (2012) and Pastore (2013) show, a problem of double information asymmetry plagues job search: from companies towards worker, of whom a company doesn’t know the skill level, on the one hand; and from the worker towards the company, of which he/she doesn’t know the real career opportunities offered, on the other hand. The consequences of the inefficient way of working of public and private operators are well Please cite this article in press as: Pastore, F. The quasi-market of employment services in Italy. Journal of Policy Modeling (2019), https://doi.org/10.1016/j.jpolmod.2019.06.008

+Model

JPO-6548; No. of Pages 22 8

ARTICLE IN PRESS F. Pastore / Journal of Policy Modeling xxx (2019) xxx–xxx

known: high (frictional and mismatch) unemployment and mass appeal to networks of relatives and friends to find a job6 , not to mention the negative consequences on inequality and social mobility. The consequence is that, on average, only 3.5% of young Italians find a job with the help of public and private Job centres in a year (Mandrone, 2011). A percentage that is clearly unsatisfactory, if compared to, say, the United Kingdom (about 7%) and Germany (13.5%) (for a more systematic comparison, see Giubileo, 2012). The impact evaluation of DUL on the regional system of employment intermediation can be made, for instance, by difference-in-difference. It involves comparison across regions between values of key variables related to a pre-reform year (for example, 2012) and a post-reform year (from 2014 onwards). This let us see if the outcome variable of interest was affected by the program and the entity of its influence. The outcome variables used in this case include: a) b) c) d) e)

the NEETS (or some subgroups) who use employment services as a job search method; the percentage of people who find a job with the help of job centers; the number of guidance, job placement and training services offered by Job centers; the percentage of beneficiaries placed over the total, and so on; the quality of services offered and the professional skills of operators involved in the program.

If the outcome variable in the post-intervention period in the region under consideration (target group) is statistically higher than in the other regions (control group), the program is deemed effective. In the case under consideration, considering that the competition principle applies to organizations of different ownership type, the evaluation exercise should include not only PES, but also forprofit and nonprofit organizations. The impact evaluation of the program on items a) and b) can be carried out by using sample survey data. It includes questions for the unemployed job seekers and those who have found a job on the actual job search method used. Items c) and d) require the use of administrative data relate to the organizations involved in the program. It would be interesting to make a comparison with the outcomes of other regions to determine if the positive change eventually occurred is due to a general policy (and, therefore, it cannot be attributed specifically to the DUL, but to any other policy (such as, for example, the Youth Guarantee) (Pastore, 2015b).7 The DUL is expected to affect also item e), that is the quality of services supplied and the professional level of operators of Job centres. Unfortunately, there are no quantitative measures on operators’ skills before and after the DUL. It is likely that this indicator was, however, also influenced by the severe economic and financial crisis, which did not allow the public and private sector to make significant new hires. Omitted results of difference-in-difference estimates based on ISFOL Plus data suggest that the DUL slightly increased the use of private, but not PES. These results are probably due to the relatively small number of recipients.

6

Nearly 45% of young people find a job this way. The Youth Guarantee (YG), for example, could increase the use of employment services and also the supply of guidance and placement services offered by Job centres all over the Country, regardless of the DUL program. 7

Please cite this article in press as: Pastore, F. The quasi-market of employment services in Italy. Journal of Policy Modeling (2019), https://doi.org/10.1016/j.jpolmod.2019.06.008

+Model

JPO-6548; No. of Pages 22

ARTICLE IN PRESS F. Pastore / Journal of Policy Modeling xxx (2019) xxx–xxx

9

3. The DUL program Profiling consists of dividing users into needy groups (fasce di bisogno), according to certain criteria that indicate the level of difficulty for beneficiaries to find a job by themselves. Following the DUL management handbook drafted by the Lombardy Region (2015), the criteria are: 1 the distance from the labor market, as measured by the unemployment duration and understood as a key measure of the autonomy level in job search; 2 education attainment level; 3 age; 4 gender. The characteristics from b) to d) are considered as corrections to point a). Each indicator is divided into sub-categories and to each sub-category is assigned a certain score8 . Profiling follows an objective method in the sense that the operator of the accredited organization taking in charge beneficiaries cannot freely decide to which group they belong to; it has simply to apply the above criteria. The group determines the maximum value of the voucher assigned and the (suggested) basket of services from which the operator, in agreement with the user, can choose to identify the individual path for outplacement (formalized in a PIP: Piano di Inserimento Personalizzato; En. Tr.: Personalized Placement Plan). Objective and automatic profiling aims to reduce the tendency to implement a creaming of users, even if creaming can still take place through the so-called “refusals”. In other words, the operator may reject beneficiaries deemed too difficult to place, pushing them to apply to other accredited organizations, typically in the public or nonprofit sector. The higher voucher for groups more in need remains, however, an important incentive against “refusal”. Objective profiling is also important to avoid a conflict of interest for the organization who does profiling and could profile a user as more in need than actual to obtain a voucher of higher amount. The first three groups contain rather homogeneous individuals, although they are characterized by an increasing level of need. The voucher allocated grows depending on the level of need. There are two types of vouchers, depending on whether they follow the employment integration path (in this case the amount of dote is from D 1950 for the group one to D 2950 for group 2, to D 3850 for group three) or the self-employment path, in which case the dote is slightly higher (from D 3720 to D 4900 to D 5875). With this dote, workers can “buy” services from accredited organizations that have taken them in charge. There is a menu of services with a list of prices that differs according to the service provided. Group four is different, as it includes employed workers who have been recently laid off and receive an unemployment benefit (cassa integrazione), who can get a fixed amount of D 2000. In this case, workers cannot start a new job until when they end the period for which they are covered by their unemployment benefit. By adhering to the program, they can buy a basket of services, including essentially vocational training activities, such as coaching, training, promotion of specific knowledge in the field of business management, tutoring and coaching to training, skills certification. The set of services offered and the relative prices is summarized in Table 2.

8 For a more detailed description of the groups in need within the DUL, see: ARIFL (2013) and Cerlini and Giubileo (2015).

Please cite this article in press as: Pastore, F. The quasi-market of employment services in Italy. Journal of Policy Modeling (2019), https://doi.org/10.1016/j.jpolmod.2019.06.008

+Model

ARTICLE IN PRESS

JPO-6548; No. of Pages 22 10

F. Pastore / Journal of Policy Modeling xxx (2019) xxx–xxx

Table 2 General scheme remuneration for Services by needy groups. Area of services A) Basic services

B) Welcoming and guidance

C) Capacity building

D) Job placement and other actions E) Other actions Total

Services Welcoming and access to services; Specialist interview; Path definition. Skill assessment; Self-employment tendencies and attitudes; Creating a support network; Guidance and training for an active job research; Continuous tutoring. Training – Promoting specific skills within business management Tutoring and coaching – Skills certification Placement Self-employment (alternative to placement) For job placement paths For self-employment

Group 1

Group 2

Group 3

Group 4









D 210

D 450

D 665



D 1000

D 1200

D 1350

D 2000

D 740

D 1300

D 1835



D 2510

D 3250

D 3860



D 1950 D 3720

D 2950 D 4900

D 3850 D 5875

D 2000 D 2000

Source: Our translation of the DUL Notice.

4. Descriptive analysis This section aims at analyzing the distribution of vouchers by group in need and some categories, such as individual characteristics, type of organization and so on9 . The data set is formed by the census of all beneficiaries of the measure. For a more detailed description, see the Methodological Appendix. Fig. 1 reports the DULs supplied by group of needs. Group 2, but above all group 3 are much more numerous than the others. Group 3, alone, includes more than 50% of beneficiaries. Added to group 2, they reach 78%. Considering that group 4 includes a very small number of beneficiaries (8.9%), with different features from the unemployed and the inactive, it is likely that there is a problem of appropriate definition of the groups. A best definition of groups may reduce the peaks noted in groups 2 and 3, therefore making more effective the DUL implementation and, at the same time, reducing some of the possible distortions. One can hypothesize the existence of a trade-off between the number of groups, freedom of choice by the beneficiaries and possible reduced effectiveness of the policy. The more and the better defined are the groups (and, thus, also the type of services for them) at the beginning, the lower the role of personal choice by accredited operators entity, the lower the possibility of making mistakes about the more effective type of action for beneficiaries. This applies if the paths are sufficiently well defined to have a diversified effectiveness depending on the type of beneficiary. One of aims of the econometric analysis will be to identify the most effective services. Fig. 2 shows the overall percentage of DULs activated in the whole period by the beneficiary education attainment. As expected, the average education level of beneficiaries is lower than the

9

Table A2.1 in Annex 2 gives further details on the composition of the sample.

Please cite this article in press as: Pastore, F. The quasi-market of employment services in Italy. Journal of Policy Modeling (2019), https://doi.org/10.1016/j.jpolmod.2019.06.008

+Model

ARTICLE IN PRESS

JPO-6548; No. of Pages 22

F. Pastore / Journal of Policy Modeling xxx (2019) xxx–xxx

11

Percent 30

40

50

51

20

27.22

12.9

0

10

8.884

10

Average need Low meed

Cassa Integrazione

High need Group of Need

Fig. 1. Share of DUL activated by group of need.

Percent 20 30

40

39.15

25.78

10

14.28 7.957 5.044 1.212

4.135 .2526

.3157

.0631

0

1.804

0 o N

le t it

. y y d s) s) s) s) er er .D ar ar re st st ar ar ar ar h d a a la m e e e e i P n c y y y y r M M o P c de -5 (3 -3 -5 el el se (4 (2 (4 ot ev ev ity l l n y w y y s t I I r r r i I Lo tle rs da da ve Ti ni on on ive U c c n U se se h h ig ig H H Educational qualification of beneficiary

Fig. 2. Share of DUL activated by educational title.

Please cite this article in press as: Pastore, F. The quasi-market of employment services in Italy. Journal of Policy Modeling (2019), https://doi.org/10.1016/j.jpolmod.2019.06.008

+Model

ARTICLE IN PRESS

JPO-6548; No. of Pages 22 12

60

F. Pastore / Journal of Policy Modeling xxx (2019) xxx–xxx

20

Percent

40

53.82

14.13 11.21

9.657

8.591

0

2.589

0

Agency State

Nonprofit Cooperative Private Consortium of private Operator's ownership

Fig. 3. Share of DUL activated by ownership type.

population average. We note, in particular, a predominance of low to high secondary education. The share with university education is low, confirming that an important return to higher education is a greater than average opportunity to find a job. Fig. 3 presents the allocation of DULs by type of operator. As expected also from the theoretical analysis, the private sector has the lion’s share of DULs: private profit (57%) and non-profit (20.6%) organizations represent the most important slice. The DULs spent in public job centres are about 10%, or, if they include the independent agencies of public operators (the so-called special agencies of the legislative decree 267/2000), the public sector gets about a quarter of the DULs allocated. Fig. 4 shows the distribution of beneficiaries by group in need and by legal nature of the operator organization taking them in charge. The private operator does take in charge a greater than average number of beneficiaries of groups 1 and 2 and lower than average of group 3 and even 4. This fact can be read also the other way around: beneficiaries more easily employable tend to apply to private operators more frequently. Overall, though, shares appear to be physiological. Moreover, groups 2 and 3 are not always easily distinguishable one from the other. A more accurate division in a larger number of groups could shed a new light also on the creaming potential issue of beneficiaries by private operators. As said, the core of DUL is to give to users the possibility to choose their best accredited operators. Once the “sovereignty of users” is restored, it is clear that all accredited operators will develop all the tools - human and professional skills – necessary for producing the most efficient employment services in the long run. Unfortunately, there are no statistical figures on the operators’ skills before and after the implementation of the program. This prevents us from assessing the DUL effect on operators’ skills. However, the severe economic and financial crisis neither allow the public nor the private sector to make significant new hires. Please cite this article in press as: Pastore, F. The quasi-market of employment services in Italy. Journal of Policy Modeling (2019), https://doi.org/10.1016/j.jpolmod.2019.06.008

60

Private

56.93

40

45.75

30.73 26.6

8.449

6.539

4.29

0

4.154

60

Nonprofit

Consortium of private

55.22

Cooperative 58.98

40

51.61

24.39

20

23.14

21.49

19.71

16.32

13.58

5.322

5.947

4.293

0

Percent

19.37

16.96

10

12 11

14 13

10 15

12 11

14 13

10 15

12 11

14 13

15

In-need group Graphs by ownership1

Fig. 4. Distribution of recipients by ownership type of the organization.

ARTICLE IN PRESS

20

21.82

+Model

JPO-6548; No. of Pages 22

Agency

58.41

F. Pastore / Journal of Policy Modeling xxx (2019) xxx–xxx 13

Please cite this article in press as: Pastore, F. The quasi-market of employment services in Italy. Journal of Policy Modeling (2019), https://doi.org/10.1016/j.jpolmod.2019.06.008

State

+Model

ARTICLE IN PRESS

JPO-6548; No. of Pages 22 14

F. Pastore / Journal of Policy Modeling xxx (2019) xxx–xxx

5. Determinants of success 5.1. Methodology This section proposes an evaluation of the so-called gross impact of the program on the probability of “successfully completing the DUL”. This expression has a specific meaning in the program: it happens when, using part or all of the voucher assigned, the beneficiary gains an employment contract (as a permanent wage employee, also in apprenticeship, or a fixed-term or temporary work) for at least six months10 or, alternatively, becomes self-employed. The statistical sample available is formed by beneficiaries belonging to the first three groups (fasce di bisogno)11 that in June 2015 concluded “administratively” the DUL12 and, therefore, the estimates should be read as a comparison between the characteristics of those who concluded the DUL “successfully” and those who have had one or more employment contracts, but of the total duration of less than six months or a job as non-wage employees (i.e: participating in partnership, project work, intermittent work, socially useful work, occasional work, training), regardless of its duration. The self-employed workers are excluded because they were a very small number. The dote may also be concluded without having used all the budget planned in the PIP. If, the dote is concluded and the desired employment outcome is not reached, the beneficiary can activate a new dote by using the resources eventually not spent for the services admitted as “repeatable” (see the regional Framework of minimum standards of employment services). The analysis compares the impact of the DUL on the probability of successfully completing the program in an empirical ultra-simplified and multivariate setting, principally aiming at comparing the different types of services used, after profiling and taking in charge, controlling for some individual variables (gender, age, educational level) and variables related to the organization that has taken in charge the beneficiary (ownership type). More analytically, the estimated pattern takes the following form:  n m Pr DUL = 1| Si , Xj = 

i=1

j=1

    exp α + ni=1 βi Si + m j=1 γj Xj     1 + exp α + ni=1 βi Si + m γ X j j j=1

(1)

where DUL is a dummy equal to 1 for the beneficiary who completes “successfully” the program; Si are the different types of services supplied; whilst Xi are the control variables. Briefly, the analysis compares elements of both the demand for DUL services, expressed by beneficiaries, and the supply of DUL services, expressed by the accredited organizations.

10 Initially, the Regional authority wanted to consider only labor contracts of one year, but this was impossible due to the expiry date of the European Social Fund with which the program was financed. Note that the 6-months period can be also totalled over more labor contracts. 11 As already noted, beneficiaries belonging to group 4, cannot subscribe employment contracts since they are still formally employed, although being laid off. 12 Note that the information relative to the dependent and several independent variables comes from mandatory communications (COB) and, for several reasons, is not available for all the beneficiaries. This leads to a loss of observations on the DUL concluded from 39588 to 29185 units.

Please cite this article in press as: Pastore, F. The quasi-market of employment services in Italy. Journal of Policy Modeling (2019), https://doi.org/10.1016/j.jpolmod.2019.06.008

+Model

JPO-6548; No. of Pages 22

ARTICLE IN PRESS F. Pastore / Journal of Policy Modeling xxx (2019) xxx–xxx

15

5.2. Results As the dependent variable is discrete, the model is estimated by Logit. The outcomes are presented in Table 3 in terms of relative risk ratios, instead of estimated coefficients.13 The latter are difficult to interpret because of their highly non-linear pattern. Relative risk ratios instead are very easy to interpret. They measure the impact of the dichotomous variable considered, relative to the baseline, on the likelihood that the program is completed. As they are ratios, the values can only be positive. In particular, if they are smaller than one, they indicate that the baseline variable is more likely than the one considered; if they are greater than one, the variable considered is more likely than the baseline. Approximately 71.7% of those who complete the DUL, get an employment contract of at least six months, with a slight predominance of men (73.1%) over women (70%). This figure which is relatively high confirms the impression of Montaletti’s study that the program was quite successful. Women have a lower probability by about 15% to complete the DUL than men who participated in the program.14 This gender gap is hard to explain in terms of education, as women in the sample are more educated than men and depends probably on a greater commitment of women in unpaid work within the family, and, indirectly, on a limited availability of reconciling tools between paid and unpaid work. The risk ratios relative to a battery of dummy variables, one for every five years of age, suggest that the probability of completing the DUL follows a non-linear growing trend for young adults and then it decreases as age increases. Teenagers have a greater opportunity to complete the DUL than people in their fifties and over. The mode of the distribution is for young adults (20–29 and 30–34 years old). Age differences are more marked for women. The “negative” effect of increasing age indicates the greater difficulty of placing older adults ceteris paribus: generally speaking, adult workers have specific knowledge in a certain job, perhaps outdated and hard to export to new jobs, but fewer general skills (low educational attainment). As for education, there is little difference between beneficiaries who have compulsory education (lower secondary education), that represent the baseline, and beneficiaries with lower than the compulsory education: the corresponding coefficients are not statistically significant. Generally, companies require more than compulsory education. The same applies to those holding a professional school diploma (vocational education of 2–3 years). Those who do not declare their education level have on average a probability to close successfully the DUL approximately 17% less than those with compulsory education. The effect is due mainly to men, who have a probability of 33% less than those who have a compulsory education. Generally, people who do not declare their education level hold low attainment. Higher secondary education (4–5 years) is associated with a greater probability of completing the DUL than compulsory education by about 41%. The effect is stronger for women than for men. The data contains no information on the type of diploma. Bachelor’s and specialized university degree ensure more opportunities to complete the DUL, confirming that finding a job more easily is a return to tertiary education. A 3-year university degree ensures a greater impact than secondary high school: 60% versus 40%.

13

Relative risk ratios are obtained by taking the exponential of coefficients of Logit model: eβ . The relative risk ratio is less than one and is equal, in particular, to 0.8498, for women. This means that a woman has about 85% of the probability of a man to complete successfully the program. 14

Please cite this article in press as: Pastore, F. The quasi-market of employment services in Italy. Journal of Policy Modeling (2019), https://doi.org/10.1016/j.jpolmod.2019.06.008

+Model

JPO-6548; No. of Pages 22 16

ARTICLE IN PRESS F. Pastore / Journal of Policy Modeling xxx (2019) xxx–xxx

Table 3 Determinants of the probability of completing the DUL program. Variable

All

Women

Men

Women Age (baseline: 50 years or more) 16–19 years 20–24 years 25–29 years 30–34 years 35–39 years 40–44 years 45–49 years Educational qualification (baseline: comulsory school, namely low secondary education), of which: Title not available No title Primary education or no title Vocational education (2–3 years) High secondary school Bachelor degree (3-years) Specialised university degree (5 years) I level Master degree II level Master degree or Ph.D. Ownership type (baseline: private), of which: PES Nonprofit organization Consortium of private organizations Cooperative organizations Other state organization AFOL 1 AFOL 2 AFOL 3 AFOL 4 Afol 5 Firm 1 Firm 2 Firm 3 Firm 4 Firm 5 Firm 6 Type of services received, of which: Continuous tutoring Analysis of attitude to entrepreneurship Skill assessment Skill certification Group coaching Personal coaching Network creation Work related learning and training LLL – Permanent training LLL – Specialization Guidance to job search Group guidance to job search Tutoring and support to training

0.8498*** 1.0391 1.0513 1.0822 1.005 0.9307 0.9676 0.9946

0.8674 0.9593 1.0221 0.9072 0.8291** 0.895 0.9647

1.1473 1.1097 1.1057 1.0737 1.0174 1.0207 1.0093

0.8392** 1.3719* 0.9194 1.0735 1.4099*** 1.6002*** 1.7766*** 1.4724 1.4653*

1.1843 1.2445 1.7618* 1.0521 1.4717*** 1.7115*** 1.9162*** 1.7839* 1.6992*

0.6733*** 1.3163 0.7576* 1.1013 1.3804*** 1.5228*** 1.6658*** 0.9284 1.2401

1.2922*** 1.2074*** 1.2262* 1.2078*** 1.2868* 0.8167* 1.3620*** 0.5069*** 1.0244 1.5355 1.3467*** 0.6308*** 0.9132 1.0524 1.3861*** 1.1566*

1.2836*** 1.3940*** 1.1762 1.2160*** 1.1777 0.8999 1.3469** 0.5471*** 1.1133 1.3827 1.3848*** 0.6915*** 1.0347 1.0326 1.3089** 1.1535

1.3346*** 1.0678 1.3408* 1.2059** 1.5093* 0.7652* 1.4107** 0.4656*** 0.9386 1.7263 1.3017*** 0.5816*** 0.8274** 1.0769 1.4529*** 1.1613

0.8040*** 0.2952*** 1.1064*** 2.5895*** 1.3200** 0.6726*** 1.1726 0.4869*** 0.4518*** 0.5614*** 0.7062*** 0.6137*** 0.4945***

0.8573*** 0.2224*** 1.1015** 1.8408*** 1.1798 0.6050*** 1.4771 0.5602 0.4426*** 0.4977*** 0.7467*** 0.6312*** 0.4883***

0.7578*** 0.3816*** 1.1115** 3.4485*** 1.5089** 0.7423*** 1.0588 0.4239** 0.4567*** 0.6414*** 0.6761*** 0.5835*** 0.5018***

Please cite this article in press as: Pastore, F. The quasi-market of employment services in Italy. Journal of Policy Modeling (2019), https://doi.org/10.1016/j.jpolmod.2019.06.008

+Model

JPO-6548; No. of Pages 22

ARTICLE IN PRESS F. Pastore / Journal of Policy Modeling xxx (2019) xxx–xxx

17

Table 3 (Continued) Variable

All

Women

Men

Constant Number of observations DUL completed (Number) DUL completed (Percentage)

3.3278*** 29066 21386 71.73%

2.7150*** 13056 9326 70.01%

3.4608*** 16009 12060 73.13%

Source: Own processing on DUL data. Note: the table reports relative risk ratios. ***,**,* 1%, 5%, and 10% significance, respectively.

Fig. 5. Relative risk ratios by ownership type. Source: own elaboration on Table 3.

Note: The bar at around 1 suggests the baseline value.

Postgraduate qualifications, such as I or II level Masters or PhD, do not give a statistically significant impact on job opportunities, although coefficients are all greater than one, probably for the strong heterogeneity within this group. This outcome probably depends on the particular type of sample of NEET. One of the aims of DUL is increasing competition among accredited organizations of different ownership type. as expected by the Biagi Law of 2003, but never implemented before. Fig. 5 reports the relative risk ratio by ownership type. Contrary to the theoretical expectations of those who believe that the private sector has a competitive advantage in the activities of job placement and vocational training, the estimates suggest that the DULs activated in PESs have a greater probability by about 30% of being completed successfully. Note that here the group used as a term of comparison are private operators of small size. There are six private operators that appear regularly among the ten accredited organizations with the greatest number of activated vouchers (names omitted). They are separated from the rest of private operators, given their larger than average size. Interestingly enough, not all of them have a relative risk ratio higher than the baseline. It marks the need to monitor the performance of individual companies in search for the most efficient and the importance, in turn, of monitoring and also implementing an external rating of the effectiveness of companies in doing their placement work. Please cite this article in press as: Pastore, F. The quasi-market of employment services in Italy. Journal of Policy Modeling (2019), https://doi.org/10.1016/j.jpolmod.2019.06.008

+Model

JPO-6548; No. of Pages 22 18

ARTICLE IN PRESS F. Pastore / Journal of Policy Modeling xxx (2019) xxx–xxx

Cooperatives seem to do better even than the baseline (by about 20%). Non-profit organizations seem to have a performance only slightly lower than public Job centres. In fact, the effect is mainly due to women who have a probability to complete successfully the DUL by 40% greater than in small private centres. Whoever applies to a non-profit organization has a higher probability, by about 20%, to complete the program than the baseline, a percentage similar to that of cooperatives. This is probably also due to favorable conditions that nonprofit organizations reserve to particular categories of workers, such as the disabled and disadvantaged workers (alcoholics, drug addicts, convicts admitted to alternative measures, etc.). Those who apply to consortia have a greater probability to complete the program than the baseline, but the coefficient is not highly significant from a statistical point of view. Also among the AFOLs (names omitted), there are important differences. Generally speaking customers seem to be aware of this performance, since the most successful are those with the highest number of vouchers activated. However, an interesting finding is that it is not always true that the organizations which have the greatest number of vouchers are also those that complete them more frequently. This suggests to give to beneficiaries more information about the actual performance of accredited organizations to increase competition and increase the overall quality of services. The quality of the information provided is an important tool to stimulate continuous improvement in the quality of services provided by all organizations in the long run. As for the type of services provided, some services are more efficient than others. As a note of caveat, note that, first, some types of services have been excluded because they are provided to all (welcoming and access to services, that is taking in charge, which is necessary to access to the service) or almost all beneficiaries (specialist interview, path definition). Other services (self-employment, promoting specific knowledge in business management) have been excluded for lack of observations. In addition, it is difficult to say which is the baseline, as the beneficiaries could also indicate different services they used. The following analysis can be very useful to reshape the basket of services provided in the program. Of course, to this end, the statistical information should be integrated with other qualitative information based on direct experience. Fig. 6 reports the relative risk ratios found in Table 3 by type of service provided. Three types of services stand out for being associated to a higher chance of successfully completing the DUL: (a) skill certification; (b) group coaching; (c) skill assessment. Creating a support network is not statistically significant. Continuous tutoring, which is very common, has a rather low impact on the opportunity of completing the DUL, by about 40%. Also the analysis of individual tendencies and attitudes seems to be associated with a low risk ratio. Interestingly, while, on the one hand, skills assessment is more effective than the average by only 10%, on the other hand, skills certification is very effective (+260%), especially for men (+6 times). This outcome should not be surprising: skills certification is used for better understanding the strengths, the skills, beneficiaries’ work capacities, the type of action they need, beyond education attainment. Especially young people are indistinguishable from each other by employers, without a certification of skills. It could be supplied also by educational institutions in the future. Coaching, that is provided only for beneficiaries belonging to group 3, may have very different effects depending on whether it is carried out individually (lower than baseline probability of completing the program) or in group (higher than the baseline). Group coaching, however, is statistically significant only for men. Different from individual coaching, group coaching allows the opportunity to exchange experiences of job search among beneficiaries with similar characPlease cite this article in press as: Pastore, F. The quasi-market of employment services in Italy. Journal of Policy Modeling (2019), https://doi.org/10.1016/j.jpolmod.2019.06.008

+Model

JPO-6548; No. of Pages 22

ARTICLE IN PRESS F. Pastore / Journal of Policy Modeling xxx (2019) xxx–xxx

Fig. 6. Relative risk ratios by type of service provided. Source: own elaboration on Table 3.

19

Note: The bar at around 1 suggests the baseline value.

teristics, which explains why it is more effective. In a way, group coaching increases the network of contacts and knowledge of the person, which cannot occur in individual coaching. The other services included in the estimates, all related to vocational training, are associated to a negative performance. This may depend on the specific characteristics of those who use it, characteristics that would be associated with a lower probability of successfully completing the DUL, or with the low quality of the training itself. In principle, both the explanations are plausible. This is a recurring doubt in all policy analysis and can only be adequately faced with appropriate econometric methods (for example, randomized experiments with target and control groups). 6. Policy discussion of the results As noted above, with some differences, the DUL system has been the model of a national reform. It is, hence, of great interest from a policy point of view to assess its advantages and shortcomings. This essay has shown that the DUL policy is an important step forward to address many of the shortcomings of the Italian system of employment services, although at some cost, such as the increased public spending associarted to the supply of the overall service, also due to the proliferation of operators and the loss of economies to scale, and the lion’s share of the private sector. In line with the recent trends of policy modelling (Ruiz Estrada & Park, 2018), a number of policy prescriptions, based n our empirical analysis, may help maximize benefits and minimize shortcomings of the policy. An important point is to guarantee the right number of accredited operators: not so big to nullify economies to scale, not so small to prevent the right degree of competition among accredited operators. The correct number would be such to reach something like to German standard of about 50–70 beneficiaries per staff, depending on the ability to reduce burocratic tasks for the staff. Besides, per se, the new organization of public services is unable to fully solve all the problems typical of the Italian PES, especially the slowness of the organization without a full digitalization Please cite this article in press as: Pastore, F. The quasi-market of employment services in Italy. Journal of Policy Modeling (2019), https://doi.org/10.1016/j.jpolmod.2019.06.008

+Model

ARTICLE IN PRESS

JPO-6548; No. of Pages 22 20

F. Pastore / Journal of Policy Modeling xxx (2019) xxx–xxx

of most part of the bureaucratic tasks, such as registration, declarations of availability to work and so on. There have been several proposals in this direction to suggest the need to introduce an adequate, nationally centralized ICT system. It is not by chance that the current populist government has decided to invest one billion euros on this objective. Moreover, and linked to the previous point is the issue of the centralization of the services that has been rejected by the vote at the constitutional referendum of December 2016. It means that employment services will remain in the jurisdiction of regional bodies and this will constrain the action of the ANPAL (National Agency of Active Labor Policies) to a small number of cases. It is necessary that still the regional bodies accept some general prescriptions to omogenize the implementation of the service over the entire territory. Furthermore, the overall cost of the service and also the current lion’s share of the private sector could be reduced at once by digitalizing burocratic tasks and by providing PES with the human, technological and financial resources that are necessary to make the public actor more efficient and effective in providing the required employment services. The reform should be complemented with a sufficient investment to improve the quality of the production capacity of the public sector. This would be crucial also for a better success of the YG and of the citizenship income promised by the current populist government. To prevent the refufals by private agencies of the hardest to place workers, it might be necessary to either impose some proportion between the different in-need groups or introduce a penalization for those operators who do not maintain the right proportion, which could be that of the overall population of users. A continuous monitoring by third parties might help increase the awareness of users of the ability of operators and allow users to make the right choice for their specific needs among the accredited operators. In other words a public rating of operators would allow reducing the asymmetric information problem which might otherwise prevent benefit recipients from making the right choice. A continuous process of evaluation of the impact of the program by different operators is key to improve the way of working of the program in the long run. Also gaming will be put under treat by systematic evaluation exercises by third parties, such as research agencies. 7. Concluding remarks This study has provided an in-depth analysis of the way of working of a recent policy experiment conducted in Lombardy of transforming the sector of employment services into a quasi-market. The aim was to revitalise a moribund sector, which proves however important in other countries. Quasi-markets are already common in the health and education sector in several countries. They imply the possibility to give to users a market power by empowering them with a voucher which allows them deciding to which accredited organization to require a given service. This freedom of choice is important to generate competition among accredited organizations, including forprofit and nonprofit private organizations. Quasi-marketa in the provision of employment services present shortcomings similar to those happening in other sectors, such as: the tendency of the private sector to have a lion’s share of the market; increasing costs and tendency of the private sector to implement a creaming or cherry picking strategy of the easiest to place users and gaming with users to obtain the voucher and the reward for successfully completing the DUL without actually completing it. A number of tools have been devised in Lombardy to prevent such distortions, based on the Anglo-Saxon experience. Quasi-markets in the sector of employment services were implemented successfully in Australia and the UK previously. Cherry picking is prevented by providing vouchPlease cite this article in press as: Pastore, F. The quasi-market of employment services in Italy. Journal of Policy Modeling (2019), https://doi.org/10.1016/j.jpolmod.2019.06.008

+Model

JPO-6548; No. of Pages 22

ARTICLE IN PRESS F. Pastore / Journal of Policy Modeling xxx (2019) xxx–xxx

21

ers of different value for individuals of different needs: higher for the most in need and hardest to place. Gaming is prevented by providing a reward for successfully completing the program only when the labor contract has been obtained for at least 6 months. The empirical analysis shows some interesting characteristics of the program, such as the phenomenon of cherry picking, although it is hard to understand whether it was due to a demand pull or a supply push, since private companies do tend to provide more efficient services and it may be that the easiest to place individuals prefer to apply to private, rather than to state organizations. The analysis of the determinants of the probability to complete the DUL suggests the existence of a gender gap and a position favourable to individuals with a higher level of education attainment, although less than the average population. Differences among organizations of different ownership type are far from trivial. The relative risk ratio is not the highest among all private employment agencies. The biggest ones tend to have a better than average performance, although not all of them. In addition, some state or nonprofit organizations seem to do better than small private agencies. The fact that the best performing companies are not always the most frequently used ones suggest that it is important to give more information about performance to users, in order to allow them making better choices. In turn, this is important for the long run effectiveness of the program. Three types of services are associated with a better performance: (a) skill certification; (b) group coaching; (c) skill assessment. This is important information for policy makers, practitioners and users of the program. Acknowledgements This essay would not have been possible without the active collaboration of the Lombardy Region and, in particular, of the following persons, that the author wishes to thank: Gianni Bocchieri, Marinella Gallo, Guido Longoni, Bruno Mercurio, Giampaolo Montaletti, Alberto Rainoldi. They contributed not only by providing the statistical data, but also by helping with the data preparation for the empirical analysis and the interpretation of results. Moreover, the author wishes also to thank Laura Ciattaglia, Giampiero Falasca, Francesco Giubileo, Fabrizio Marra de Scisciolo and Roberta Piano for useful comments and insights in various steps of the elaboration of the paper as well as participants to a presentation at the Lombardy Region (October 2015). Nevertheless, the responsibility of the opinion expressed and of any shortcomings, errors or omissions is only of the author. Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.jpolmod.2019.06.008. References Annicchiarico, B., Di Dio, F., & Felici, F. (2013). Structural reforms and the potential effects on the Italian economy. Journal of Policy Modelling, 35(1), 88–109. ARIFL. (2013). La Dote Unica del Lavoro. Il sistema delle fasce ad intensità di aiuto. Descrizione della metodologia. Newsletter n. 2. October Arrow, K. (1963). Uncertainty and the welfare economics of medical care. American Economic Review, LIII(5), 941–973. Bartlett, W., & Le Grand, J. (1993). Quasi-markets and social policy. Palgrave Macmillan. Bocchieri, G. (2015). L’occasione persa dell’assegno di ricollocazione. , 19 June. www.lavoce.info

Please cite this article in press as: Pastore, F. The quasi-market of employment services in Italy. Journal of Policy Modeling (2019), https://doi.org/10.1016/j.jpolmod.2019.06.008

+Model

JPO-6548; No. of Pages 22 22

ARTICLE IN PRESS F. Pastore / Journal of Policy Modeling xxx (2019) xxx–xxx

Centeno, C., & Stewart, J. (2013). The concept of employability with a specific focus on young people, older workers and migrants, JRC technical report. Luxembourg: European Union. Cerlini, S., & Giubileo, F. (2015). Il modello Dote Unica del Lavoro in Lombardia: Un esempio di innovazione per il sistema Italia. , 27 July. www.workmagazine.it Giubileo, F. (2012). La via inglese al collocamento. , 9 March. www.lavoce.info Giubileo, F., & Pastore, F. (2013). Quale futuro per i centri per l’impiego. , 25 October. www.lavoce.info Giubileo, F., & Pastore, F. (2013). Centri per l’impiego. Il pubblico non basta. , 21 November. www.lavoce.info Giubileo, F., Leonardi, M., & Pastore, E. F. (2013). Centri per l’impiego. Ecco cosa si può salvare, Le nuvole del lavoro (Blog del Corriere della Sera) 25 November. Larsen, C. A., & Vesan, P. (2012). Why public employment services always fail. Double-sided asymmetric information and the placement of low-skill workers in six European countries. Public Administration, 90(2), 466–479. Le Grand, J. (1991). Quasi-markets and social policy. Economic Journal, 101(408), 1256–1267. Leonardi, M. (2015). Così cambiano le politiche attive del lavoro. , 1 July. www.lavoce.info Lombardy Region. (2015). Manuale di gestione della Dote Unica. Milano. Mandrone, E. (2011). La ricerca del lavoro in Italia: l’intermediazione pubblica, privata e informale. Politica Economica, 27(1), 83–124. Mauro, L. (2004). The macroeconomics of Italy: A regional perspective. Journal of Policy Modelling, 26(8–9), 927–944. Montaletti, G. (2015). Un modello controfattuale per la valutazione delle politiche del lavoro. Settembre: Newsletter ARIFL, n. 15. Pastore, F. (2013). Employment services in view of the school-to-Work transition. A comparative analysis. Rome: FORMEZ. Pastore, F. (2015a). The youth experience gap. Explaining national differences in the school-to-work transition, springer briefs. Heidelberg: Springer. Pastore, F. (2015b). The European youth guarantee: Labor market context, conditions and opportunities in Italy. IZA Journal of European Labor Studies, 4(May (11)). Pastore, F. (2019). Why so slow? The school-to-work transition in Italy. Studies in Higher Education, 44(8), 1358–1371. http://dx.doi.org/10.1080/03075079.2018.1437722 Ruiz Estrada, M. A., & Park, D. (2018). The past, present and future of policy modeling. Journal of Policy Modeling, 40(1), 1–15. Weishaupt, J. T. (2011). Social partners and the governance of public employment services: Trends and experiences from Western Europe. ILO. WD n. 17, May.

Please cite this article in press as: Pastore, F. The quasi-market of employment services in Italy. Journal of Policy Modeling (2019), https://doi.org/10.1016/j.jpolmod.2019.06.008