HIRING INCENTIVES AND/OR FIRING COST REDUCTION? EVALUATING THE IMPACT OF THE 2015 POLICIES ON THE ITALIAN LABOUR

MARKET∗

Paolo Sestito♦ and Eliana Viviano♦

(Preliminary and incomplete: December 2015)

Abstract

Aiming at reducing labour market dualism and fostering job creation, the Italian government has recently adopted two closely related policy measures: a quite generous incentive for firms hiring workers with permanent job contracts and, as part of a wider reform package known as the Jobs Act, a reduction of firing costs for newly established permanent contracts. In this paper, using timely administrative microdata on hiring and firing occurred in Veneto, a large region in the North-eastern part of Italy, we evaluate the separate impact of the two interventions. Using a diff-in-diff approach and some differences in the design of the two policies, we find that both policies were quite successful in improving individuals’ probability of finding a stable job and increasing firms’ employment.

JEL: J6, C93 Keywords: job creation, firing costs, hiring incentives, labour market reforms.

∗ We thank Bruno Anastasia and Gianluca Emireni of Veneto Lavoro for providing us not only with the data but also valuable help in interpreting them. We also thank Effrossyni Adamopoulou, Clemence Berson, Emanuele Ciani, Emanuela Ciapanna, Marta de Philippis, Guido de Blasio, Sebastien Roux and Francesco Zollino for helpful comments. We are solely responsible for all errors. The views expressed herein are ours and do not necessarily reflect those of the Bank of Italy. ♦ Bank of Italy, DG Economics, Statistics and Research.

1

1. Introduction

Italy lost 1 million jobs over the 2008-2014 period as a result of a double deep recession.

As a consequence of its dualistic labour market structure, job losses were concentrated among

youths and more generally people holding temporary contract positions. More recently,

employment and more specifically open ended contracts have started to rise, albeit the still

subdued GDP pick-up (grown in 2015 a bit less than 1%).

In this paper we analyse the role in such an evolution of two policy measures adopted in

Italy at the end of 2014 and aimed at both reducing labour market dualism and stimulating job

creation. The first is a sizable temporary rebate of non-wage labour costs which applies to all

new permanent job contracts (hereafter PHI) of workers who in the previous semester had not

an open-ended position. The incentives are not targeted to specific groups of workers nor

conditional to firm-level net job creation and apply also to conversions from a fixed-term to an

open-ended position. The second measure is a reshaping of firing regulations, aimed at reducing

the uncertainty and the average expected costs of firing for all new permanent contracts in firms

with at least 15 employees (the “contratto a tutele crescenti”, hereafter CTC, part of a wider

reform package denominated “Jobs Act”).

We separately identify the effects of the two policy interventions by jointly exploiting the

time dimension (one having occurred since January and the other since March 2015) and the

differences in the coverage of the two schemes as the PHI applies to all firms but only to some

workers (those with no permanent job position over the previous semester), while the new CTC

regulations reshaped firing costs only for firms beyond the 15 employees threshold. In order to

do that, we exploit the administrative data for the Veneto region, which allows us not only to

measure the different labour market flows in the first half of 2015 (and 2 years before), but also

to reconstruct the previous labour market status of workers, to identify the firm matched with

the workers and the firm’s size.

It has to be stressed that there are relevant aspects of the two policies here not considered.

We do not discuss the pros and cons of the design of the two policies, nor the general

equilibrium effects that can derive from their implementation.

Even with these caveats, our results, however, show clearly that both policies fostered net

job creation at the firm level and increased the workers’ (permanent) job finding probability. In

the period January-June 2015 in Veneto the total (net) flow of new job positions increased by 40

per cent. According to our estimates hiring incentives alone contributed by 12 percentage points

2

and the combination of the two policies added other 10 percentage points. We also find that,

among quite homogeneous subsamples, also the CTC alone stimulated firms’ hiring.

The two policies also contributed to substantially increase the rate of conversion of fixed-

term jobs into permanent positions. A broader finding of our paper is that the PHI may have

affected positively both permanent and temporary contract hiring, as firms have resorted to the

latter for workers unknown to them, in order to test their skills before eventually proposing a

permanent contract (the eligibility condition being related to the absence of a permanent

contract over the semester before the permanent contract hiring or conversion).

One of the contributions of our paper is to provide direct evidence that a reduction in the

uncertainty and expected costs of firing increases (short term) job creation at the firm level. This

is relevant for two reasons. Firstly, because the firing costs in the pre-reform regime were

mostly those related to the high uncertainty stemming from the possibility that judges had to

decide in favour of the worker and mandating the worker’s reinstatement in the firm. Second,

because the literature about the labour demand effects of a certainty equivalent change in firing

costs is by itself a bit ambiguous. From a theoretical point of view, in a static model firing costs

are just a tax on firing that are economically equivalent to a component of labour costs. If wages

cannot adjust, an increase in firing costs implies a downward shift in labour demand and lower

employment levels; its relevance may be however contained by the fact that firing costs may be

far away over time and needs to be discounted. In a dynamic context, higher employment

protection reduces the extent to which firms adjust labour input to economic shocks, with

negative effect on the allocation of resources. While the theoretical literature on the effects of

firing costs on employment and allocative efficiency is well developed, the empirical literature

is relatively rather scant. Autor et al. (2007) using state-level and time variation in employment

protection legislation in the US find that higher employment protection reduced employment

fluctuations. Adhvaryu et al. (2013) focus on rural India and relate supply-side shocks (like

rainfalls) and different state-level employment protection legislation to show that employment

protection reduces employment responses to shocks. In our paper we instead focus on a single

Italian region, before and after the inception of the new law on firing costs. As a consequence,

our results are not affected by spurious local trends.

Concerning hiring incentives, to the best of our knowledge, this is the first time that the

Italian government introduces non-targeted, non-conditional hiring incentives. Cipollone and

Guelfi (2003) analyse selective hiring incentives introduced in 2001 targeted to young workers

hired on a permanent basis and do not find relevant effects on labour demand. Ciani and De

3

Blasio (2014) consider instead a very short-term policy intervention for the conversion of fixed-

term contracts into permanent jobs, introduced in Italy in 2013 and lasting just few weeks

because of severe funding constraints. This incentive was targeted to females and young people

and they find a positive effect of the policy on conversion rates. In 2014 the government

introduced incentives to firms hiring workers on a permanent basis, but incentives were

conditional to firms’ net job creation. Given the very weak economic conditions, only very few

firms applied for the incentives. Our results, instead, support the hypothesis of the effectiveness

of non-targeted non-conditional incentives on net job creation, and in this respect are similar in

spirit to the ones analysed by Cahuc et al. (2014) who find a positive and rapid expansionary

effect of the non-conditional hiring credits introduced in France during Global Financial crisis.

Our results are also in line with Neumark (2013) and Neumark and Grijalva (2013) who argue

that non-targeted hiring incentives have a positive effect on employment during recessions, even

if deadweight losses associated to these policies are generally large (e.g. Brown 2011).

The paper is organized as follows. In section 2 we briefly describe the policy measures

introduced by the Italian government in 2015 (hiring incentives and the Jobs Act). In section 3

we describe our dataset. In section 4 we present our results on both net job creation and on the

individual probability to find a job. Section 5 briefly present some robustness checks. Section 6

concludes.

2. The measures under scrutiny

The Italian labour market is heavily segmented in permanent and fixed-term workers. The

dualism arose at the end of the nineties when the government progressively introduced different

types of fixed-term contracts, to increase flexibility in the use of labour. Higher flexibility,

however, was not accompanied by changes in firing costs for permanent job contracts.

During the 2000s the share of fixed-term workers in total workers increased rapidly to

around 13 per cent. More than 60 percent of hires were made up of fixed-term job contract,

often used also as a cheap screening device before hiring workers under a permanent contract.

People employed temporarily suffered the most for the consequences of the Global financial

crisis, when firms, facing a sudden drop in their activity, used all the available margins to adjust

labour input. For this reason the first steps in tackling firing costs were made in 2012, by the

law no. 92/2012, the so-called “Fornero Reform”, followed in 2015 by the Jobs Act.

4

Before that, Italy was characterised by firing costs whose main feature was its uncertain

and potentially quite high amount1. As such, dismissals were not only possible but totally

costless for the firm if taking place on a just cause basis, i.e. workers misbehaviour or firms’

need to reduce or reorganize its workforce2. However, whenever a worker opposes and the court

judges the dismissal to be unfair, the costs may become rather high in the case of firms with

more than 15 employees. Firms with less than 15 employees could opt (according to the law no.

108/1990) between reinstating the worker and paying a pre-set severance payment (related to

workers seniority and varying from 2.5 to 6 monthly pay). For firms with at least 15 employees,

the general rule was that of workers’ reinstatement. On practical grounds firms could side-step

reinstatement by reaching private arrangement with the individual worker, the cost of it possibly

being quite high.

The most critical aspects of this regime were due to the uncertainty in both the timing and

contents of judges’ deliberations, whose inclinations are deemed to have varied quite a lot

across both time and space3. The widespread delays in the Italian civil justice system are also

thought to have compounded the inefficiencies as workers were lacking any income support

(and pay) pending the litigation, while firms could end up having to pay quite high wage arrears

if finally succumbing.

Most of the analysis made in Italy about the implications of this regime considered the

immediate effect of the regime discontinuity around the 15 employees threshold. There is

evidence that the potentially higher costs affecting firms with 15+ employees may have

somehow refrained firms’ growth when firms happened to approach the threshold (see

Schivardi and Torrini, 2008). The effects, while statistically significant, were not economically

much relevant, as the discontinuity in the firms’ size distribution around the threshold is not

much marked, vis-a-vis for instance the sharp discontinuity taking place in France around the 50

employees threshold relevant for the rules concerning the role of unions at the firm level (see

e.g. Garicano, Lelarge and Van Reenen, 2013; Gourio and Roys, 2014). Such a limited impact

may however be due to measurement error in measuring firms’ size, as actually there were ways

1 We here refer to individual dismissals. 2 A liquidity cost is present because the firm has to pay to any departing worker (including those quitting and those retiring) a given percentage of each yearly earnings accumulated over time at a subdued interest rate (more precisely it is equal to ¾ of the CPI inflation rate plus 1.5%), the so called Trattamento di Fine Rapporto (TFR) . Disbursing the TFR may imply an additional firing cost at the margins for those firms who are liquidity constrained as such an anticipated disbursement has to be somehow financed through a banking loan or other means . Notice that since 2007 workers may decide to accumulate the yearly contributions to the TFR into a second pillar pension scheme, depriving the firm of the low cost financing provided by TFR: in such a case, firing a worker is totally unaffected by the existence of the TFR. 3 See Ichino (1996).

5

to circumvent the stringency of the threshold, for instance by employing apprentices or other

temporary contract workers whose presence did not count in order to define the applicability of

the threshold. Moreover, the general equilibrium implications of such a restraint to firms’

growth might have been wider than what appears around the relevant threshold.

So, as a matter of fact removing the risk of these potentially high firing costs has been an

hotly debated issue in Italy for over 20 years. Some attempts were made by both the first

D’Alema government (in 2000) and, more prominently, the second Berlusconi government (in

2002). The latter’s policy initiative was blocked by the strong reactions of the biggest union, the

CGIL, exploiting the art 18 of law 300/1970 as a successful flagship.

Both the Fornero Reform (law no.92/2012) and more recently the Jobs Act (decree no.

183/ 2014) were aimed at reducing the costs of uncertainty in firms above the 15 employees

threshold, while leaving fair dismissals costless. The Fornero law attempted to better fix the

existing procedures: litigants were channelled into a conciliatory procedure so as to prevent

often lengthy judicial litigations, restrained the possibility of reinstatement in the case of unfair

dismissals for disciplinary or economic reasons and provided an upper bound to the monetary

firing costs whenever these had to be applied. However, the reform was considered too timid

because of the room left to the judges in determining both the fairness of the dismissal and its

consequences for firms. The second attempt, i.e. the Jobs Act, was more forcefully, possibly

because the government decided not to buy in unions’ consent. The Jobs Act further restrained

the possibility of workers reinstatement, limiting it to discriminatory dismissals and to some

very specific cases of disciplinary dismissals, and dictating as a general rule that unfair

dismissals have to be compensated by disbursing an amount of money. Such an amount is

predetermined by the law as it increases with job tenure (from a minimum of 4 times the

monthly pay to a maximum of 24 times, i.e. 2 monthly pay every year of seniority).

Furthermore, such a monetary compensation may be halved if the worker accepts a transaction,

so ending any pending litigation about the nature of the dismissal. To foster transactions for a

temporary period (up to 2018), the acceptance of such transaction is cashed tax free by the

worker.

Differently from the Fornero reform, covering the stock of permanent employees, the

newly established rules apply only to the permanent contracts signed after March, 7th, 2015,

when the new law came into force. The new law applies also to all open-ended contracts signed

by firms with more than 15 employees (or by those which by the new hires would be

overcoming the 15+ threshold). Existing employees (as well as the newly signed permanent

6

contracts in firms staying below the 15 employees threshold) remain covered by the previous

law arrangements.

All in all the new regime reduces both the expected firing cost and the amount of

uncertainty surrounding it for firms over the 15 threshold, with no change for those below the

threshold4. The amount of such a reduction is however not easy to be assessed for at least two

reasons. First, a relevant component is the reduction in the perceived uncertainty, which may

vary a lot across firms, as, even in the case of risk neutral firms, the uncertainty-related costs

depended, in the past, on the behaviour of local courts. On top of that one has to consider that

the tax free transaction option has two different possible implications: on the one hand it

reduces the cost for a firm of a dismissal deemed as unfair by the judge; on the other hand it

may stimulate the worker to resist (and the judge to deem as unfair) a dismissal whose fairness

is rather uncertain, as in case of no resistance (and no unfairness decision) the worker would get

nothing.

The government in the meantime introduced a very generous non-conditional hiring

incentive. The incentive, established by the Financial Stability Law for 2015 (but already

announced at the end of October 2014) was covering all new permanent workers hired by any

firm from January to December 2015, provided the worker did not have a permanent contract in

the previous 6 months. The incentive is a three-year exemption from social security

contributions up to a threshold which is quite high compared with the average contributions

typically paid by firms to workers (according to the government estimates the incentive should

fully cover the social security contributions of almost 80 percent of new hires). Notice that also

conversions from fixed-term to permanent job contract within a given firm are subsidized

(conversions from apprenticeship are instead excluded as the apprenticeship contract is already

subsidized).

The two policy measures undertaken in 2015 and analysed in this paper almost overlap,

because both target permanent hires and job contracts conversions from fixed-term into open-

ended. There is however a slight difference in their timing: from January 2015 for the incentive,

from March 7th, 2015 for new firing costs. Moreover, some differences in the population

4 To be more precise, for those approaching the 15 employees threshold there is a reduction in the costs implied by overcoming the threshold as they are exempted from the implications of overpassing it. This is why, differently from Schivardi and Torrini (2008) we are considering not only firms close to the threshold in analysing the possible effects of the reform. For firms close to the threshold (say firms with 14 employees) the relevant discontinuity would be that with firms also below but much farther away from the threshold (say firms with 1 to 5 employees), for which no change was implied by the new law. However, focusing upon such a comparison would have implied looking at a much smaller subsample.

7

targeted by the two policies can be used to separately identify their effects. Incentives are paid

to firms of any size while the Jobs Act applies to firms with at least 15 employees; the

incentives apply only to workers with no permanent contract in the previous 6 months, while the

previous contract status of the worker is irrelevant for the application of the Jobs Act. Thus,

information about firm size and workers' past work histories, together with the precise date of

the new contract, allow for separate identification of the effect of the two policies.

While leaving to the next sections a more detailed discussion of our identification

strategy, it has to be noticed that the many reasons behind firms’ use of temporary contract and

the fixed amount of the hiring incentive for all new permanent contracts signed in 2015 leave a

lot of room for firms’ “strategic” behavior. A firm may cash in the same amount for both a

permanent contract signed in January 2015 and one signed in December 2015. This means that,

insofar as the worker is well known and positively valued by the firm - for instance because he

or she was holding a temporary contract position in the same firm - the firm may have an

incentive to offer immediately a permanent contract cashing in the incentive and avoiding the

risk that some other firm offers the worker such a chance removing her from the pool of

workers eligible for the incentive. Whenever the worker is unknown to the firm, and insofar

December 2015 is not yet arrived, the firm may opt for a lower risk path, as the worker may be

hired on a temporary basis - such a temporary contract not destroying the hiring incentive

eligibility status of the worker and allowing a test of her skills - converting later the fixed-term

position into an open-ended one.

This implies that the hiring incentive PHI may boost not only permanent hiring (and

conversions from fixed term to open ended contracts) but also temporary contract hiring,

especially in those cases where testing workers’ skills is relevant. This implies that the effect of

the two policy measures introduced in 2015 may not be easily obtained by comparing the trend

in the number of newly signed open-ended contracts (CTC) and new temporary job contract5.

This issue, however, is more extensively discussed in section 4.

5 The creation of temporary job contracts might be affected insofar as there are fraudulent arrangements between the worker and the hiring firm in order to “build up” the eligibility status of the worker by passing through a temporary contract. While - as shown by Veneto Lavoro (2015) – clearly fraudulent cases are present in the data, we believe that the quantitative amount of the extreme case here described is contained by the risk that the worker would have to bear by renouncing to a permanent contract in order to get the promise to obtain a new permanent contract in a different firm in 6-month time. As a matter of fact (see section 4) the job to job moves flow has remained quite unchanged over time.

8

3. The data and some evidence

In this paper we use administrative microdata about the so-called Comunicazioni

Obbligatorie. In Italy all occurrences concerning a job position must be electronically notified

to the Regional agencies in charge of active labour market policies (and also made accessible to

the Italian social security institute, INPS). Microdata archives, which cover only employees in

the private sector and part of the public sector, are collected and organized by each Italian

region with only a few aspects more recently available in a similar fashion over the whole

country. Potentially, the database registers when the position is created, destroyed, converted

from fixed-term to permanent, or whether the duration of a fixed-term contract is extended.

Because of the decentralization of the data-collection process, the quality of the micro data

differ quite a lot across the different regions. Our data refer to one Italian region, Veneto,

characterized by high quality and high timeliness. Veneto is located in the North-Eastern part of

the country. The weight of manufacturing in total economy is among the highest in the country,

as well as labour market participation. On average, the weight of this region in total dependent

employment is around 8 per cent (excluding agriculture and Public Administration).

Our dataset contains information regarding all events (hiring, firing, conversion and

prolonged duration) occurred in Veneto between January 2013 and June 2015 (we also use

some information concerning 2012 in order to better construct the workers’ previous history).

For each event it is possible to uniquely identify the firm and the worker. On top of the relevant

anonymized (firm and worker) identifiers, we know the firm’s size6 (by size class) and sector of

the activity and the worker’s gender, birth-date, nationality. For each event we know the type of

job contract, the place of work and the relevant dates (day, month, year). So, we may

reconstruct each firm’s and individual’s work histories.

From our dataset we exclude job transitions of domestic workers hired by households,

public sector workers and those in the agricultural sector, as they are not subject to the policy

measures. After this selection our dataset includes around 2,3 million episodes involving

800,000 workers and almost 200,000 firms.

In this paper we use three dimensions of our dataset. First, we consider data on monthly

net job creation at the aggregate level. Net job creation is defined as the difference between

newly created positions (hiring) in a given month and the number of positions destroyed in the

6 The size of the firm is not directly communicated at the time of hiring/firing/conversion, but can be recovered by other compulsory reporting required by the law about legally protected workers (Law 68/1999).

9

same month. We consider all types of job destruction, i.e. firing, workers' voluntary separation,

retirement.

We select both permanent and fixed-term contracts events. The latter group includes: (i)

standard fixed-term dependent contracts, (ii) agency workers, (iii) apprentices7, (iv)

collaboration workers (so-called parasubordinati, i.e. a sort of consultants employed on a

temporary basis) and (v) internships (so called tirocini-formativi). For each month from January

2013 to June 2015 we calculate the cell-level net flow, distinct by type of job contract (the 6

types mentioned above), firm size (less than 15 employees, 15+) and workers’ past working

status, i.e. whether the worker was/was not employed with a permanent job contract during the

previous 6 months. In what follows, for simplicity we label the latter workers as “eligible for the

incentive”, independently on the year in which we observe them (i.e. even if we observe them in

2013 and 2014, when no incentive was in place for them).

For computing cell-level net hiring, we need to define also whether the worker involved

in a job separation is eligible. We conventionally define as “eligible” those who are separated

from a fixed-term job contract, and as “non-eligible” those separated from a permanent

position8. This definition is adopted also to check whether firms’ substituted workers already

employed with an open-ended job contract with other workers eligible for the hiring incentive.

Second, for some estimates, we consider the firm-level panel dimension implicit in our

dataset, i.e. the monthly flow originated by each firm in our dataset. For each firm we separately

consider the flow of hires with permanent and temporary job contracts by eligibility status. We

randomly select a closed panel of 5,000 firms followed from January 2013 to June 2015. Thus,

each month for each firm in the sample we know the total amount of hires by type of job

contract and workers’ past-working status (employed/non-employed with a permanent job

contract in the previous 6 months).

Last, we also use the individual-level panel dimension of our dataset, recording the

working status of individuals month by month. We randomly select 50,000 individuals among

those with at least one occurrence from January 2013 to April 2015. Thus, each month for each

individual we know whether he/she is not or is working, and the contract type.

7 To be more precise, since 2008 apprenticeship in Italy are considered by the law as open-ended contracts (they were previously considered as fixed-term). However, since no firing cost is associated to them we classify them as fixed-term. 8 In principle, a worker exiting from a temporary contract might not be personally eligible if she has held a permanent contract during the previous semester but before the now ending temporary contract. We look more precisely to the eligibility status of the workers now hired, as this is crucial in setting whether she is eligible to the PHI, while adopting the above mentioned more approximate criterion when establishing whether the separation belong to an “eligible” cell.

10

Figure 1 plots hires and aggregate net job creation, from January 2013 to June 2015, for

both open-ended and fixed-term job contracts. In January 2015 both the number of new open-

ended and fixed-term contracts increased considerably, and then declined, even if open-ended

contracts remained at a level remarkably higher than the one registered in the corresponding

periods of 2013 and 2014. The total net flow of newly created job contracts was positive for

both types of contracts (bottom panel), even if at the end of the period it was positive for fixed-

term job contracts only.

In 2015 also conversions from fixed-term job position to open-ended positions increased

considerably, as shown by Figure 2. They increased both in absolute terms (dashed line) and

relative to the number of fixed-term positions (solid line), peaking in April 2015.

As already mentioned in Section 2, firms have the possibility to hire a worker on a fixed-

term basis, test her skills and then convert the contract into an open-ended (still benefitting for

the incentives and the Jobs Act). This two-step strategy is in principle always convenient for the

firm whenever the worker has not a better alternative (and insofar as the PHI does not decreases

or vanish, as after December 2015). The only risk for the firm stems from the fact that the

worker may have a better alternative and, if offered a permanent contract by another firm, may

disappear from the pool of eligible workers. Given that we are analysing a single region and a

relatively short period of time, we are not attempting to take account of these “local labour

market” effects. What we may however consider is the fact that, ceteris paribus, the propensity

to use such a two-step strategy is larger whenever the firm does not know the worker’s

characteristics. We proxy such a feature by distinguishing between two types of matches

between workers and firms: those involving workers who were already employed in the same

firm before (we label this group as “known workers”) and those who are matched with a firm

for the first time (workers “unknown to the firm”). The upper part of Figure 3 plots the flow of

new job contracts, by worker-firm relationship. The solid line refers to contracts involving

workers and firms who were never matched before; the dashed line refers to matches involving

workers already known to the firm. The flow of new hires of known workers increased

considerably in January 2015, when incentives were introduced and then declined. The flow of

new hires involving unknown workers peaked instead in March and increased also after April

2015. The bottom part of Figure 3 plots the share of known workers in total hires of permanent

workers and confirms that after the introduction of the PHI in January 2015 firms preferred to

hires workers whose skills were already tested in the past.

11

As a first step, we first check differences in trends before and after the inception of the

two policies, by comparing eligible vs. non eligible workers (in Figure 4) and of firms of

different size (in Figure 5), independently of the job contract type. Even if the comparisons are

affected by monthly seasonality, in Figure 4 the difference in the trends of eligible and non-

eligible workers is never significant before January 2015 and then differs afterwards. In Figure

5, instead, we do not find clear-cut evidence of a change in trends after the inception of the Jobs

Act in March 2015.

Notice however that the presence of the above depicted two-step strategy implies that the

number (or the share) of new hiring (or conversion from fixed-term contracts) into permanent

positions is not a sufficient statistics to be looked at as the policy interventions might have

boosted also temporary hiring. To identify the effects of the two policies, we need to identify a

“control” or non-treated group of events totally unaffected by them. This means focusing upon

events in small firms (as such unaffected by the CTC) by workers who were neither

immediately nor prospectively eligible for the PHI. People belonging to such control group are

the workers transiting from a permanent in a small firm to another permanent contract in the

same firm class-size. Their job-to-job moves are driven by the materialisation of a better match,

mostly because of idiosyncratic reasons unrelated to any labour demand shifts related to the PHI

or the CTC. The evolution of these flows might be due to business cycle features, as a more

buoyant economy normally triggers them. But this is precisely what one would like to control

for by comparing the evolution of the other flows, potentially affected by the two policies, and

the control group evolution.

Thus, in Figure 6 we report the flow of hiring of non-eligible workers in small firms and

the difference between the trends in some other relevant flows and the control group flow. More

in detail, panel (a) of Figure 6 reports the difference in trends of eligible and non-eligible

workers in small firms. Panel (b) reports the difference in trends of small and large firms in the

group of eligible workers. Panel (c) and (d) respectively report and compares the trend of non-

eligible hires in small firms and all other hires (of eligible workers in both small and large firms

and of both eligible and non-eligible workers in large firms). In panels (a) and (d) the existence

of a difference in trends emerges clearly after January 2015. In panel (b), where the effect of the

Jobs Act is calculated by controlling for eligibility, together with some effect probably due to

seasonality, there is also significant evidence of a change in trends in the last months of the

period, i.e. after March 2015.

12

4. The estimation strategy

The charts in previous sections show that while there are discontinuities in the

employment and hiring flow trends along some of the lines affected by the two policy measures

under consideration, the picture is very complex as the policy measures may have impacted

more than one labour market flow. In particular, the two-step strategy of offering a temporary

position later on to be transformed into a permanent contract – so as to test the worker’s skill

and still cashing in the whole amount of the PHI – may have boosted also the temporary

contract hiring. Broadly speaking, a possible overall measure of the counterfactual business

cycle evolution unrelated to the two measures may be obtained by focusing upon those labour

market flows unaffected by the two measures, permanent hiring of non-eligible workers in small

firms, a flow largely explained by permanent contract job-to-job moves within the small firms

segment.

In order to get a more precise and quantitative assessment of the effects of the two

policies we may estimate a diff-in-diff model in which the evolution of specific labour market

variables (gross flows or the change in employment) is the dependent variable. As said, we may

look at those phenomena at three different levels, the individual worker, the individual firm or

the aggregate of a labor market segment. The nature of the phenomena that we may look at may

differ across these levels, as, for instance, in the case of the individual worker we may recollect

the job finding and the fixed-term to open-ended contract conversion rate only for those

individuals we may observe, who are those who had at least one hiring, conversion or separation

event since 20139. Similarly for employment we do not observe its level, but only the gross

flows (hiring) and the net change in total employment (and employment according to different

contract categories) in individual firms and their aggregates.

The least controversial exercise looks just at conversions of fixed term into open-ended

contracts. This is a flow directly affected by both policies, although along different lines,

depending upon the workers’ previous history, defining the eligibility to the PHI, and the firm

size, relevant for the CTC. More specifically, we will look at the number of these conversions at

the cell level and at the probability of such a conversion for all temporary workers “at risk” of

being converted.

9 We do not observe people who have never get an employment (the ones remaining unemployed or out of the labor force) and those who, employed over a permanent contract, have never changed their status since 2011.

13

The cells will be given by the intersection of 30 months, 2 firm class sizes (below and

above the 15 employees threshold) and 2 worker’s eligibility status (depending upon the

worker’s history during the previous semester). At the individual level we will look at the

probability, for a worker who is holding either an apprentice or a standard fixed-term contract,

to have her contract converted into an open-ended contract in that same firm10. More formally

we estimate:

[1] ngwym = γg + γw + γy + γm +

+βD(w=1)(y≥2015) + δD(g=15+)(y≥2015)(m≥March) + ϵgwym

where γg, γw, γy, γm are respectively fixed effects for firm’s size-class (g), for worker’s

eligibility (w), for year (y) and month (m). The variable D(w=1)(y≥2015) is a dummy equal to 1 if

the conversion occurs from January 2015 onwards and involved workers eligible for the hiring

incentive. The dummy D(g=15+)(y≥2015)(m≥March) is a dummy equal to 1 if hires occur after the

inception of the Jobs Act, in a 15+ firm. The year and month dummies take account of what

happens to the conversion flows (at the cell level) and the conversion probability (at the

individual level) as a consequence of seasonality and business cycle effects. In other words, the

control group refers the conversions taking place to non-eligible workers in small firms. Notice

that in the individual level estimates we may even control for individual fixed effect.

The second exercise looks at permanent gross hiring, which is also a flow affected by the

two policies and along different lines (depending upon the workers’ previous history, defining

the eligibility to the PHI, and the firm size, relevant for the CTC). Again, the evolution of the

same flow in small firms and for non-eligible workers may provide for the relevant control

group. The estimates will be again conducted at two different levels, one of which is the

individual worker (we look at the probability for an individual to get such a permanent contract)

and the other is the individual firm. In the case of the individual probability of finding a

permanent job, the people “at risk” include both the temporary workers above considered and

all people currently jobless (but who have had at least 1 hiring or separation since 2011). Notice

that when considering individuals we may only recover the effects of the PHI11, while at the

10 Also at the cell level we consider only the conversions stemming from apprentices and standard fixed-term contracts taking place in a given firm. 11 Actually, the job finding rate is the sum of two events, finding a job in a small firm and finding it in a large firm and what one could look at are these two components, one affected only by the PHI and the other one affected also by the CTC.

14

firm level we may distinguish the effects of the two policies. More formally, for the firm level

analysis we estimate:

[2] ngofwym = γg + γo + γf + γw + γy + γm +

+𝛽𝛽1D(o)(w=1)(y≥2015) + 𝛿𝛿1D(g=15+)(o)(y≥2015)(m≥March) + ϵofgwym

where the indices 𝑜𝑜 and 𝑓𝑓 indicate respectively open-ended and fixed-term contracts and γo and

γf are the corresponding fixed-effects. The variable D(o)(w=1)(y≥2015) is a dummy equal to 1 if

the permanent hiring occurred from January 2015 on, and involved eligible workers (and 0

otherwise). The dummy D(g=15+)(o)(y≥2015)(m≥March) is a dummy equal to 1 if the permanent

hiring occurred after the inception of the Jobs Act, in a 15+ firm (independently on eligibility).

In equation [2] 𝛽𝛽1 and 𝛿𝛿1 identify the direct effect of the two policies. As before, moth and year

dummies capture what is happening because of business cycle and seasonal factors to the

permanent hiring of non-eligible workers in small firms, a flow largely explained by permanent

contract job-to-job moves within the small firms labour market segment. Notice that in the

equation we may also control for firm’s fixed effect, γf, disaggregating the class size dummy as

such used for identifying the CTC effects. Further, in such exercise we may also control for the

distinction between workers who are already know to the firm and those who are not, so as to

verify whether the PHI, and the CTC as well, have add different effects among the two

categories.

The individual level job finding probability equation is rather similar but for the absence

of the firm and the CTC effects (as well as the distinction between those already know and those

unknown by a specific firm, as these characteristics can be defined only in the case of a job

match). What we can however add is the control for individual worker fixed effects. Fully

exploiting the worker’s past work history, we can also split the sample among those who are not

employed at a given time and search for a job and those who have a fixed-term job.

The last and final exercise tries to get together the firm’s and worker’s sides of the market

by looking at the change in the employment stock. We look here at the level of 720 cells defined

by the intersection of 30 months, 2 firm class sizes (below and above the 15 employees

threshold), 2 worker’s eligibility status (depending upon the worker’s history during the

previous semester) and 6 contractual types (open-ended contracts, standard fixed-term

dependent contracts, agency workers, apprentices, collaboration workers, i.e. the so-called

15

parasubordinati, and internships, i.e the so called tirocini-formativi). Formally the estimating

equation is:

[3] ngofwym = γg + γo + γf + γw + γy + γm +

+𝛽𝛽1D(o)(w=1)(y≥2015) + 𝛽𝛽2D(f)(w=1)(y≥2015)

+ 𝛿𝛿1D(g=15+)(o)(y≥2015)(m≥March)+𝛿𝛿2D(g=15+)(f)(y≥2015)(m≥March) + ϵofgwym

The variable D(o)(w=1)(y≥2015) is a dummy equal to 1 if the permanent hiring occurred

from January 2015 on, and involved eligible workers (and 0 otherwise). The variable

D(f)(w=1)(y≥2015) is equal to 1 if instead the job contract was a fixed-term one and to 0

otherwise. The dummy D(g=15+)(o)(y≥2015)(m≥March) is a dummy equal to 1 if the permanent

hiring occurred after the inception of the Jobs Act, in a 15+ firm (independently on eligibility);

the dummy D(g=15+)(f)(y≥2015)(m≥March) refers to fixed-term hires occurring in large firms after

the inception of the Jobs Act. In equation [2] 𝛽𝛽1 and 𝛿𝛿1 identify the direct effect of the two

policies, while the terms 𝛽𝛽2 and 𝛿𝛿2 capture substitution or complementarities induced by the

policies on other types of contracts.

5. The results

5.1 Temporary to open-ended contract conversions

We start in Table 1 with conversions from fixed-term to open-ended contracts at the

aggregate cell level. Each cell ngwym refers to the number of conversion from fixed-term and

apprenticeship to open-ended contract. The first column refers to conversion of non-eligible

workers (so the cells under exam are 60 out of the total 120) and it is aimed at identifying the

effect of the CTC by comparing cells pertaining to firms above and below the 15 employees

threshold. The effect appears to be positive and statistically significant. The second column

includes also the cells measuring the conversions involving workers who have had a permanent

contract in the previous semester or an apprenticeship contract (whose conversion does not

allow to cash in the PHI). It allows to identify separately the effect of the CTC and PHI and

their interaction. While positive, the CTC now is far from being statistically significant, while

both the PHI and the interaction terms are positive and statistically significant (the other main

coefficients are reported in Table A1 in the appendix).

Table 2 moves us to the individual worker level conversions. Notice that here we may

also control for individual worker fixed effects as many people have had several temporary

16

employment spells (and practically all of them have had the chance of being converted to a

permanent contract over several months). The first two columns refer to all conversions, the last

column on the effect of conversion among non-eligible, to identify the effect of the Jobs Act

upon a more homogenous sub-set of fixed-term contract episodes, those who do not satisfy the

PHI eligibility rule12. As in the previous aggregate level table, the CTC has a sizable impact

when considered in this more restricted sub-sample of non-eligible workers, while as such its

coefficient is positive but not statistically significant when considered the larger sample. The

interaction between CTC and PHI has on the contrary a sizable and statistically significant

impact as the own effect of PHI, which turns out strongly positive. Notice that the effects are

all quite sizable when compared to the relevant average size of the monthly probabilities of

conversion, reported in the last row of the table.

5.2 Permanent contract gross hiring

We now look at permanent hiring. As said, we look at this issue from both the firm’s

(Table 3) and the worker’s (Table 4) sides. Table 3 considers permanent contract hiring at the

firm level and estimates the following version of equation [2]:

[2’] nigofwym = γi + γg + γo + γf + γw + γy + γm +

+𝛽𝛽1D(o)(w=1)(y≥2015) + 𝛽𝛽2D(f)(w=1)(y≥2015)

+ 𝛿𝛿1D(g=15+)(o)(y≥2015)(m≥March)+𝛿𝛿2D(g=15+)(f)(y≥2015)(m≥March) + ϵiofgwym

where γi indicates firm i-th fixed effects. The inclusion of firm fixed effects allows us to control

for average firm-level hiring, which can depend on unobservable firm-specific characteristics.

In some specifications we also split the dummies D(o)(w=1)(y≥2015) and

D(g=15+)(o)(y≥2015)(m≥March) into two groups to identify the effect of the two policies on

workers already “known” to the firm, i.e. those for which there is lower uncertainty about the

goodness of the job match. The results of this exercise are reported in Table 3. As before we

report the coefficients regarding permanent hiring only, while other coefficients are reported in

Table A2 in the appendix. Notice that the two policies effects are obtained by using as a control

group the hiring flow of non-eligible workers in small firms, as captured by month and year

dummies (as well as the firms’ fixed effects).

The first column reports the effect of incentives only, which, is positive and highly

significant. The second column splits the effect of incentive by type of worker (“known” vs.

12 Remember that we reconstructed such a rule also for the 203 and 2014 years.

17

“unknown”). PHI boosted the chance for both workers, but, as expected on the basis of our

discussion of the two steps strategy of exploiting the PHI, the increase is larger for the “known”

ones13. Column 3 includes the effect of the CTC which by itself is not statistically significant,

while matters quite a lot when interacted with the PHI. The last column reports the same

estimates by distinguishing also the effect of the CTC among known and unknown workers. The

new contract appear to have boosted the permanent hiring of those workers already known by

the firm, while decreasing those of the workers unknown to the firm. The result is a bit puzzling

as the new CTC, a permanent contract with reduced firing costs, would be expected to increase

the propensity of firms to take the risk of hiring, on a permanent basis, an unknown worker

more than it has shifted the propensity of hiring an already known worker. However, it has to be

noted that large firms might have less problems in finding acceptance by workers to go through

a testing period as a temporary worker and might have a larger pool of suited known candidates

before experiencing the need to risk taking on a permanent basis an unknown worker, an option

which is less costly than before but still more costly than the alternative two steps strategy

before depicted.

Notice that all these estimates also include dummy for anticipation effects, capturing the

fact that both measures were known a couple of months before their being operative (the PHI

was sketched and publicised by the government at the end of October 2014 and the new CTC

issued at the end of December 204). There is evidence (albeit some individual coefficients are

not statistically significant) that firms have timed their permanent hiring in order to exploit the

entrance of both the PHI and the CTC. These and all the other coefficients are reported in Table

A2 in the appendix.

On the basis of these estimates, the last rows of Table 3 report the size of the estimated

impact upon these flows of the two policies.

Table 4 turns to the worker’s side, by looking at the her probability of finding a

permanent job. More precisely we look at the probability that unemployed or fixed-term

workers find a permanent job position, given their characteristics. We first define a variable

πpwym which is equal to one if the worker p-th is hired with a permanent job contract in year y

and month m and is equal to zero otherwise (in case of no employment or if the job contract is

fixed-term). The other indices are defined as in the previous sections. We the estimate the

following model:

13 The coefficients of the interaction term between “known workers” and contract types are almost always positive, with the exception of internship, a quite obvious result because by law a worker vannot be hired with an internship contract more than once (see table A2).

18

[3] πpwym = γp + γw + γy + γm + βD(w=1)(y≥2015) + ϵpwym

where γp are individual fixed effects, aimed at capturing workers’ unobserved heterogeneity.

Since we have now a panel of individuals and the variable πpwym typically does not

change after a person has found a permanent job, after this event we drop the worker from the

sample (unless she is again separated from her job and re-enter the pool of job seekers, i.e. those

“at risk” to find a permanent position).14

Of course, given that the size of the firm is available only for workers finding a job,

equation [3] allows us to identify only the effects of incentives. The first column of Table 4

refers to all individuals (i.e. non-working or working under a fixed-term job contract), Columns

2 and 3 report the same estimates distinguishing workers by their past work experience, to

identify the effect of incentives also on the flow from unemployment into employment. The

results confirm the positive effect of incentives, which is extremely large compared with the

average probabilities observed in 2013/14.

5.3 The employment dynamics

We now look at net job creation at the aggregate level and we use a cell –level dataset,

where each cell is identified by the year, the month, the type of the contract, the eligibility

condition of the worker involved in the match (as before, whether the worker was not employed

with an open-ended contract in the 6 months preceding the new job) and the size of the matched

firm. Following equation [2], ngofwym now represents the number of jobs created in each cell.

In this context, equation [2] can be viewed as a decomposition of the newly created jobs plotted

in Figure 1. The estimated coefficients, reported in Table 5, represent the increase in the average

size of the cell, due to the policies. The Table for simplicity reports only the direct effect of the

policies on open-ended contracts, i.e. the coefficients 𝛽𝛽1 and 𝛿𝛿1. The other coefficients are

reported in Table A3, in the appendix.

The results of the first column indicate a positive impact of hiring incentives. The second

column controls for possible effects of anticipation of the policy measure, which could have

reduced net job creation in the last quarter of 2014, as firms could have strategically postponed

hiring in the first months of 2015 to get the incentives. The effect of anticipation is captured by

a dummy equal to 1 for flows of eligible workers employed with open-ended contract, from the

time of the announcement of the policy, in October 2014 to its implementation in January 2015.

14 A more suitable, but rather complex estimation model could be a panel duration model.

19

The coefficient of this dummy is positive but not significant and, more importantly, its inclusion

does not affect the main results.15 The last column of the Table includes also the direct effect of

the Jobs Act (the coefficient of the dummy D(g=15+)(o)(y≥2015)(m≥March)), and its interaction

with hiring incentives (as well as dummies to capture possible anticipation of the two

measures16). While the main effect of the Jobs Act is positive but not statistically significant, the

effect of the interaction term is positive and significant.

To assess the quantitative relevance of these results we look at the net flow percentage

increase observed in the first semester of 2015. In that period the number new open-ended job

positions increased by around 40% with respect to the average of the corresponding periods of

2013 and 2014. According to the estimates presented in the last column of Table 1, hiring

incentives contributed to the growth by 12 percentage points, while the joint implementation of

hiring incentive and the Jobs Act contributed by 10 percentage points.

Of course it is not possible to make credible aggregate estimates for Italy as a whole, as

both the underlying trends and composition effects may differ by geographical area. However,

under the simplifying assumption that firms located in different regions reacted similarly to the

two policies, our estimates suggest that the two policies contributed by around 0.3 percentage

points17 to the 0.5 percent increase in total private-sector permanent employment observed in

Italy in the first semester of 2015 (0.15 pp the contribution of incentives only, 0.15 pp the

weight of the interaction of the two policies; around 45,000 additional permanent workers).

6. Robustness checks

We have carried out several robustness checks. First, since also the apprenticeship job

contract has been subject to several legislative changes during the period under analysis, we

have carried out the same regressions presented in Table 3 and Table 5, but excluding

apprentices. Results, available upon request, are unaffected.

Second, to further check whether the positive effect of the two reforms on total net flows

is not due to changes of workers from temporary job positions in one type of firms to permanent

job positions another type of firm, we look at the probability that workers in small firms find a

job in large firms as a consequence of hiring incentives and vice versa. We select the pool of

15 The sample size of the first two columns is half of the sample size of the third column as we do not distinguish cells by size of firms, as we do are not estimating the effect of the Jobs Act. 16 Anticipation of the Jobs Act is captured by a dummy equal to one for net job creation in large firms and with permanent job contracts occurring from the announcement of the law in December 2014 to the inception of the Law, in March 2015. 17 Average of the two scenarios.

20

individuals as in equation [3]. We first define a variable equal to 1 in the case of a transition

from a fixed-term job in a large firm to a permanent position in a small one. We then look also

at the opposite transition. The results, available upon request, show that even if the incentives

had a positive impact on the probability of observing these transitions, the quantitative effect is

very small and, more importantly, remarkably lower than the one reported in the first column of

Table 4. As an additional check, in some estimates we exclude from the sample those firms

around the 15-employees threshold, as their behaviour might be impacted by the CTC through

additional channels. Also in this case the results are qualitatively similar.

7. Conclusions

In this paper, using a diff-in-diff strategy, we analyse the reaction of firms to two policies

introduced in the first part of 2015 by the Italian government, aimed at both reducing labour

market dualism and favouring job creation. The first is a generous hiring incentive to firms

offering open-ended job contracts, not conditioned to firms’ net job creation. The second is the

reduction of firing costs for firms with at least 15 employees (not only a monetary reduction, but

also a decrease in uncertainty about the consequences of unfair dismissals). We find that the two

policies were successful in both reducing dualism and stimulating labour demand, even during a

recession period characterised by very high macroeconomic uncertainty.

As already said, our estimates do not consider all the relevant aspects relevant in judging

the appropriateness of the measures undertaken (see Brown et al., 2011, for a wider theoretical

discussion on hiring subsidies). In particular, we do not discuss the pros and cons of the current

temporary and rather unselective hiring incentives vis-à-vis the host of often cumbersome but

permanent and more selective subsidies targeting specific groups of workers supposed to be

weaker and less employable (mostly youths and long term job seekers). Furthermore, we do not

deal with the merits and pitfalls of having incentives for gross hiring so that also the normal

turnover taking place in a firm is incentivized insofar as people are hired through a permanent

contract. The combination of the two elements implies that the policy favours the conversion of

temporary contract into permanent ones and possibly the poaching of suitable temporary

workers from one firm to another. So, an indirect effect of the hiring incentives might be that of

lifting up the temporary hiring of people whose contract is later on transformed into a

permanent one. At the moment we can only testify both an increase of the probability to find a

permanent position for both non-employed and temporary workers in other firms, and an

increase in the temporary-to-permanent contract conversions within the same firm.

21

An additional element to be considered is the temporary nature of the hiring incentives.

Part of its effects are likely to be due only to temporal substitution effects: hiring now (in 2015)

an applicant gives a cost rebate lasting until December 2017, such a rebate being fully lost when

postponing the hiring of the same applicant to 2016. Part of the effects that we find in the first

months of 2015 might be due to the postponement of hiring from the late moths of 2014 to the

very beginning of 2015, that our robustness check can only partially rule out, as we cannot

control for more complex strategic behaviour of firms.

Our estimates also fall short of an overall evaluation of the new firing rules introduced by

the Jobs Act. As a matter of fact, the Jobs Act has left unchanged the general principle that only

dismissals opposed by the worker and considered unfair by a judge have to be compensated.

Differently from the previous regime, the reinstatement of the worker, while still possible,

applies only to few and better specified cases of unfair dismissals (the ones deemed to be

discriminatory). Also the uncertainty concerning the amount of the financial compensation

possibly stemming from a judiciary intervention has been considerably lowered as its amount

has been capped and pre-specified by the law as an increasing function of worker seniority.

Furthermore, the compensation cost for firms may be halved if the worker accepts a transaction,

so ending any pending litigation, whose acceptance is favoured as the compensation so obtained

is cashed tax free. The new regime is likely to provide more certainty to both the worker and the

firm. At least for the time being, i.e. the three years over which the above mentioned tax

exemption has been granted, it may induce an higher share of dismissals to end up with some

compensation (the risks of high compensation costs has been reduced, but the firm may still find

more convenient to avoid any risk at all by offering the tax exempt transaction). Our estimates

do not allow to consider all these shifts, whose relevance will increase over time as the stock of

employees will be increasingly made up by people hired according to the new rules. Neither we

are able to consider to what extent the tax exemption is relevant and the possible effects of its

removal in the future.

Furthermore, our estimates do not consider the overall general equilibrium implications

of the reduced uncertainty and average amount of the dismissal costs for the firm. We do find

that firms reacted to these changes18 and even small effects above the 15 employees threshold

18 The existing literature did not find clear cut evidence of a sharp discontinuity around the 15 employees threshold over which the previous dismissal regulations entered into action. Schivardi and Torrini (2008) find evidence of a statistical significant effect over the firm’s growth around the threshold, as firms tried to avoid overpassing the threshold. However such an impact was of no much economic relevance (as evidenced by the lack of large spikes in the firms’ size distribution). This may be partly due to the fact that there were ways to circumvent the threshold binding, for instance by hiring apprentices and temporary workers, who were exempt from the general dismissal

22

might have significant general equilibrium effects, as they may cumulate to other market

imperfections (for instance capital market imperfections) in shifting the whole firms’ size

distribution.

All in all, our estimates are therefore only one of the elements necessary to decide what to

do in the future, concerning the possible presence of selective work incentives (i.e. incentives

targeted to population groups considered less easily employable) and the tradeoffs between

marginal tax wedge reductions (i.e. tax cuts applied only to either gross or net hiring) and tax

wedge reductions applied across the board (i.e. inward shifts in the tax schedule facing the

whole stock of workers).

We however believe that our empirical exercise is a step forward for the comprehension of the

reaction of firms to changes in employment protection and in its interaction with changes in

other labour cost components.

rules. To the extent that these rules-avoidance mechanisms had negative effects upon firms’ efficiency, they were not irrelevant from a general equilibrium point of view.

23

References Adhvaryu A., Chari A. V., and Sharma S. (2013) “Firing costs and flexibility: evidence from firms’ employment responses to shocks in India” The Review of Economics and Statistics, 95(3): 725–740. Autor D., Kerr W. and Kugler A. (2007) “Does Employment Protection Reduce Productivity? Evidence From US States”, The Economic Journal, 117(521), 189–217.

Brown, A. J.G., C. Merkl and D. J. Snower, (2011), “Comparing the effectiveness of employment subsidies” Labour Economics, 18, 168-179. Cahuc P., S. Carcillo and T. Le Barbanchon (2014) “Do Hiring Credits Work in Recessions? Evidence from France”, IZA Discussion Paper No. 8330. Ciani and de Blasio (2015) “Getting Stable: An Evaluation of the Incentives for Permanent Contracts in Italy”, IZA Journal of European Labor Studies, 4:6 Cipollone P. and A. Guelfi (2003) “Tax credit policy and firms’ behaviour: the case of subsidies to open-end labour contracts in Italy” Bank of Italy, wp. no. 471. Garicano, L., C. Lelarge, and J. Van Reenen (2013), “Firm size distortions and the productivity distribution: Evidence from France.” NBER working paper no. 18841 Gourio F. and N. Roys (2014) “Size-dependent regulations, firm size distribution, and reallocation” Quantitative Economics, 5, 377–416. Ichino, P. (1996), Il lavoro e il mercato, Mondadori Neumark, David. 2013. “Spurring Job Creation in Response to Severe Recessions: Reconsidering Hiring Credits.” Journal of Policy Analysis & Management, Vol. 32, No. 1, pp. 142-71. Neumark, D. and Grijalva, D. (2013) The Employment Effects of State Hiring Credits During and After the Great Recession, NBER Working Papers 18928. Schivardi, F. and R. Torrini, (2008) "Identifying the effects of firing restrictions through size-contingent differences in regulation," Labour Economics, Elsevier, vol. 15(3), pages 482-511, June. Veneto Lavoro (2015) “Le assunzioni sospette: decontribuzione e comportamenti opportunistici delle imprese” n.65/2015

24

Tables and Figures Figure 1: Hiring and net job creation by type of job contract (thousands).

2030

4050

fixed

-term

05

1015

open

-end

ed

Jan-2013 Jan-2014 Jan-2015time

open-ended fixed-term

Hiring

-40

-20

020

fixed

-term

-10

-50

510

open

-end

ed

Jan-2013 Jan-2014 Jan-2015time

open-ended fixed-term

Net job creation

25

Figure 2: Number of conversions from fixed-term or apprenticeship job contracts to permanent job contract, and ratio to the total number of hiring with fixed-term or apprenticeship job contracts.

0

1020

30ra

tio

020

0040

0060

0080

00co

nver

sion

s

Jan-2013 Jan-2014 Jan-2015time

conversions ratio

26

Figure 3 Hiring (all types of contracts) by previous relationship with the worker: known to the firm (i.e. employed in the past in the same firm) and unknown to the firm (never employed with the firm) in the upper part; share of known workers in total permanent hires in the bottom panel.

1000

015

000

2000

02500

030

000

3500

0U

nkno

wn

50001

000015

0002000

025

000

3000

0Kn

own

Jan-2013 Jan-2014 Jan-2015time

Known Unknown

Flow of known and unknown w.

.15

.2.2

5.3

.35

know

n in

ope

n-en

ded

Jan-2013 Jan-2014 Jan-2015time

Share workers

27

Figure 4: Hiring (all types of contracts) by previous employment condition: employed with an open-ended contract in the previous 6 months (non-eligible) and unemployed or employed with a fixed-term job contract in the previous 6 months (eligible). Thousands in the upper panel and difference between the two trends and confidence intervals (small vertical lines) in the bottom panel. differences are normalized by using the pre-treatment average.

Note: robust standard errors.

24

68

10N

on-e

ligib

le

1020

3040

50El

igib

le

Jan-2013 Jan-2014 Jan-2015time

Eligible Non-eligible

Hiring

-.005

0.0

05.0

1

Jan-2013 Jan-2014 Jan-2015time

lower/upper Difference

Diff. in trends

28

Figure 5: Hiring (all types of contracts) by size of the firm: less than 15 (not subject to the Jobs Act) and 15+ (subject to the Jobs Act). Thousands in the upper panel and difference between the two trends and corresponding confidence intervals (small vertical lines) in the bottom panel. differences are normalized by using the pre-treatment average.

Note: robust standard errors.

10

1520

2530

<15

1015

2025

3015

+

Jan-2013 Jan-2014 Jan-2015time

15+ <15

Hiring

-.004-

.002

0.0

02.0

04

Jan-2013 Jan-2014 Jan-2015time

lower/upper Difference

Diff. in trends

29

Figure 6: Flows (hiring) and difference in trends between hires of non-eligible workers in small firms and other hires. Differences are normalized by using the pre-treatment average.

Note: robust standard errors.

-.005

0.0

05.0

1.01

5.02

Jan-2013 Jan-2014 Jan-2015time

lower/upper Difference

(a) Small firms: elig. vs. non-eli

-.004-

.002

0.0

02.0

04.0

06

Jan-2013 Jan-2014 Jan-2015time

lower/upper Difference

(b) Eligible: large vs. small firms

2030

4050

60ot

hers

02.

55

7.5

10N

on-e

lig.,

<15

Jan-2013 Jan-2014 Jan-2015time

Non-elig., <15 others

(c) Flow non-elig small firms vs. others

-.005

0.0

05.0

1

Jan-2013 Jan-2014 Jan-2015time

lower/upper Difference

(d) Diff: Non-elig. small firms vs. others

30

Figure 7: Flows (hiring) and difference in trends between hires of permanent workers and other hires (other eligible in panel (a); other employed in large firms in panel (b); non-eligible in small firms in panel c). Differences are normalized by using the pre-treatment average.

Note: robust standard errors.

-.01

0.0

1.0

2.0

3

Jan-2013 Jan-2014 Jan-2015time

lower/upper Difference

Permanent in small firms: eligible workers vs. others eligible

-.01

0.0

1.0

2.0

3

Jan-2013 Jan-2014 Jan-2015time

lower/upper Difference

Permanent and eligible: large and small firms

24

68

1012

othe

rs

02.

55

7.5

10N

on-e

lig.,

<15

Jan-2013 Jan-2014 Jan-2015time

Non-elig., <15 others

(c) Permanent: flow non-elig small firms vs. others-.0

10

.01

.02

.03

Jan-2013 Jan-2014 Jan-2015time

lower/upper Difference

All permanent vs. others

31

Table 1: Aggregate estimates of the effect of the policies on conversions from fixed-term to permanent job contracts

Non eligible All

Jobs Act 76 136

(0.065)* (0.127)

Incentives

749

(0.000)***

Incentives*Jobs Act

993

(0.000)***

Observations 60 120 R-squared 0.853 0.882 Increase in total flow (post/pre) 103% 63% Estimated effect (contribution in percentage points): Jobs Act 34pp Incentives 35pp Incentives*Jobs Act 26pp R-squared 0.853 0.882 Robust p values in parentheses. * significant at 10%; ** significant at 5%; *** significant at 1%. The dependent variable is the number of fixed term contract converted into open ended contracts in each cell level. Each cell is defined by the intersection of month (from January 2013 to June 2015), firm size (below/above 15 employees) and worker’s eligibility to PHI. In the first column only the cells pertaining to non- eligible workers are considered.

32

Table 2: Individual-level estimates of the effect of the policies on the probability of conversion from fixed-term to permanent job contract

All Non-eligible

(1) (2) (3)

Incentives 0.013 0.015

(0.000)*** (0.000)***

Jobs Act

0.002 0.004

(0.369) (0.000)***

Incentives*Jobs Act

0.019

(0.000)***

Individual fixed-effects yes yes yes Year and month dummies yes yes yes All direct effects and all interactions yes yes yes Observations 359,102 359,102 42,173 Monthly probability of conversion in 2013-14 .004 .004 .001 Robust p values in parentheses. * significant at 10%; ** significant at 5%; *** significant at 1%. The dependent variable is a dummy equal to 1 if the fixed-term job contract is converted into an open-ended contract. In the last column only the cells pertaining to non- eligible workers are considered.

33

Table 3: Permanent contract hiring: firm-level estimates of the effect of the policies

(1) (2) (3) (4) Incentives 0.017 0.017

(0.000)*** (0.000)*** Jobs Act

0.000

(0.663) Incentives*Jobs Act

0.015

(0.000)*** Incentives * Unknown

0.008

0.008

(0.000)***

(0.000)***

Incentives*Known

0.013

0.013

(0.000)***

(0.000)***

Jobs Act*Known

0.016

(0.000)***

Jobs Act*Unknown

-0.017

(0.000)***

Incentives*Jobs Act

0.023

(0.000)***

Firms' fixed effects

yes yes yes Year and month dummies

yes yes yes

All direct effects and all interactions

yes yes yes

Observations 6,131,520 6,131,520 6,131,520 6,131,520 Increase in total flow (2015/14) 34% 34% 34% 34% Estimated effect (contribution in percentage points): Incentives 24 24 Jobs Act 0 0 Incentives*Jobs Act 2 2 Incentive * Known 14 14 Incentive * Unknown 10 10 Jobs Act* Known 2 Jobs Act * Unknown -2 Robust p values in parentheses. * significant at 10%; ** significant at 5%; *** significant at 1%. The dependent variable is the number of hires made by each firm. For each firm we identify cells measuring the number of hires of workers with given characteristics made by the firm. Each cell is defined by the intersection of month (from January 2013 to June 2015), type of contract (6 categories), eligibility to PHI (as defined in the text).

34

Table 4: Permanent employment: individual-level estimates of the effect of the policies on the probability to find a permanent job

All Non-working Fixed-term

(1) (2) (3)

Incentives 0.008 0.008 0.011

(0.000)*** (0.000)*** (0.001)***

Individual fixed-effects yes yes yes Year and month dummies yes yes yes All direct effects and all interactions yes yes yes Observations 1,289,075 860,185 428,890 Monthly probability to find an open-ended job position in 2013-14 .008 .008 .005 Robust p values in parentheses. * significant at 10%; ** significant at 5%; *** significant at 1%. The dependent variable is a dummy equal to 1 if the person finds a permanent job position and 0 otherwise. The second column refer to people moving from unemployment into a permanent employment; column (3) refers to transitions from fixed-term to open-ended job contracts in another firm within the same month.

35

Table 5: Aggregate estimates of the effect of the policies on net job creation

(1) (2) (3)

Incentives 7648 7786 4936

(0.063)* (0.060)* (0.002)***

Jobs Act

105

(0.854)

Incentives*Jobs Act

3995

(0.019)**

Anticipating incentives

549 262

(0.512) (0.457)

Anticipating incentives in large firms

26

(0.965)

Anticipating Jobs Act

-263

(0.721)

Direct effects and all interactions yes yes yes Observations 360 360 720 R-squared 0.675 0.675 0.637 Increase in total flow (post/pre) 40% 40% 40% Estimated effect (contribution in percentage points): --Incentives 19pp 20pp 12pp --Incentives*Jobs Act 10pp Robust p values in parentheses. * significant at 10%; ** significant at 5%; *** significant at 1%. The dependent variable is the net employment growth at each cell level. Each cell is defined by the intersection of month (from January 2013 to June 2015), type of contract (6 categories), eligibility to PHI (as defined in the text). In the first two columns only the cells pertaining to firms with less than 15 employees are considered. In the third column the observations include also the cells relative to the firms with 15+ employees.

36

Appendix

Table A1: Main interactions not reported in Table 1

(1) (2)

Coeff. p-value Coeff. p-value

Large 121 0.001 20 0.562

Eligible

197 0.001

Large#Eligible

92 0.300

Standard fixed-term 74 0.021 The models also include year and month dummies and the interactions between size, eligibility, type of contract (standard-fixed term or apprentices), interacted also with year dummies.

37

Table A2: Main interactions not reported in Table 3

1

2

3

4

Coeff. p-value Coeff. p-value Coeff. p-value Coeff. p-value

Anticipating incentives -0.002 0.000 -0.002 0.000 -0.001 0.010 -0.001 0.010 Anticipating incentives in large firms

-0.006 0.000 -0.006 0.001

Anticipating Jobs Act

0.001 0.398 -0.005 0.000 Eligible 0.008 0.000 0.008 0.000 0.008 0.000 0.008 0.000 Large firms 0.000 0.562 0.000 0.562 0.000 0.414 0.000 0.616 Eligible*Large firms 0.001 0.002 0.001 0.000 0.001 0.000 0.001 0.000 Eligible*year2015*Apprentices 0.003 0.000 0.003 0.000 0.003 0.000 0.003 0.000 Eligible*year2015*Standard fixed-term contracts 0.009 0.000 0.009 0.000 0.009 0.000 0.009 0.000 Eligible*year2015*Agency 0.000 0.605 0.000 0.605 0.000 0.434 0.000 0.432 Eligible*year2015*Collaborators 0.003 0.000 0.003 0.000 0.003 0.000 0.003 0.000 Eligible*year2015*Intern 0.000 0.758 0.000 0.758 0.000 0.610 0.000 0.608 Post*Eligible*year2015*Apprentices -0.002 0.001 -0.002 0.001 -0.002 0.001 -0.002 0.001 Post*Eligible*year2015*Standard fixed-term 0.000 0.910 0.000 0.910 -0.002 0.136 -0.002 0.136 Post*Eligible*year2015*Agency 0.001 0.083 0.001 0.083 0.001 0.060 0.001 0.059 Post*Eligible*year2015*Collaborators -0.005 0.000 -0.005 0.000 -0.005 0.000 -0.005 0.000 Post*Eligible*year2015*Intern 0.003 0.000 0.003 0.000 0.003 0.000 0.003 0.000 Large*Eligible*year2015*Apprentices -0.001 0.432 -0.001 0.432 -0.002 0.393 -0.002 0.391 Large*Eligible*year2015*Standard fixed-term 0.006 0.103 0.006 0.103 0.006 0.110 0.006 0.110 Large*Eligible*year2015*Agency 0.004 0.076 0.004 0.076 0.004 0.086 0.004 0.087 Large*Eligible*year2015*Collaborators -0.005 0.003 -0.005 0.003 -0.005 0.002 -0.005 0.002 Large*Eligible*year2015*Intern 0.006 0.003 0.006 0.003 0.006 0.004 0.006 0.004 Post*Large*Eligible*year2015*Apprentices -0.003 0.174 -0.003 0.174 -0.003 0.174 -0.003 0.174 Post*Large*Eligible*year2015*Standard fixed-t. -0.015 0.001 -0.015 0.001 -0.015 0.001 -0.015 0.001 Post*Large*Eligible*year2015*Agency -0.004 0.140 -0.004 0.140 -0.004 0.140 -0.004 0.140 Post*Large*Eligible*year2015*Collaborators 0.004 0.022 0.004 0.022 0.004 0.022 0.004 0.022 Post*Large*Eligible*year2015*Intern -0.003 0.309 -0.003 0.309 -0.003 0.309 -0.003 0.309 Known -0.006 0.000 -0.006 0.000 -0.006 0.000 -0.006 0.000 Small*Eligible*Known -0.002 0.000 -0.003 0.000 -0.002 0.000 -0.005 0.001 Large*Non-eligible*Known -0.001 0.000 -0.001 0.000 -0.001 0.000 -0.002 0.001 Large*Eligible*Known -0.004 0.000 -0.004 0.000 -0.004 0.000 -0.009 0.001

Unknown*Permanent -0.001 0.004 -0.001 0.080 -0.006 0.001 -0.003 0.000

Known*Permanent 0.003 0.000 0.002 0.000 -0.002 0.001 -0.001 0.000

Unknown*Apprentices -0.005 0.000 -0.005 0.000 -0.009 0.001 -0.005 0.000

Known*Apprentices 0.001 0.000 0.001 0.000 -0.003 0.001 -0.003 0.001

Unknown*Standard Fixed Term 0.019 0.000 0.019 0.000 0.020 0.001 0.020 0.001

Known*Standard Fixed Term 0.021 0.000 0.021 0.000 0.022 0.001 0.022 0.001

Unknown*Agency -0.005 0.000 -0.005 0.000 -0.005 0.000 -0.005 0.000

Known*Agency 0.001 0.001 0.001 0.001 0.001 0.000 0.001 0.000

Unknown*Collaborator -0.005 0.000 -0.005 0.000 -0.006 0.000 -0.006 0.000

Known*Collaborator 0.002 0.000 0.002 0.000 0.002 0.000 0.002 0.000 The models also include year and month dummies as well as all interactions between year (2013 and 2014), type of contract, the eligibility status and the size of the firm. The variable post is a dummy equal to 1 for observations from March 2015 on.

38

Table A3: Main interactions not reported in Table 5

1

2

3

Coeff.

p-value Coeff.

p-value Coeff.

p-value

Eligible -23 0.964 -23 0.964 20 0.940

Large firms

2273 0.000

Eligible*Large firms

-2326 0.000

Apprentices -2889 0.016 -2889 0.016 -3449 0.000

Standard fixed term contracts -893 0.441 -893 0.443 -2392 0.000

Agency workers -2350 0.048 -2350 0.049 -3347 0.000

Collaborators -2861 0.017 -2861 0.017 -3417 0.000

Intern -2922 0.001 -2922 0.001 -3484 0.000

Open-ended*Small firms

525 0.082

Open-ended*Eligible -8582 0.000 -8582 0.000 -5423 0.000

Apprentices*Eligible -278 0.691 -278 0.692 -366 0.282

Standard fixed term*Eligible -3560 0.01 -3560 0.01 -2469 0.001

Agency*Eligible -1327 0.18 -1327 0.181 -357 0.291

Collaborators*Eligible -529 0.464 -529 0.466 -372 0.38

Eligible*year2015*Apprentices 166 0.801 166 0.802 513 0.069

Eligible*year2015*Standard fixed-term contracts 2461 0.187 2461 0.187 2609 0.003

Eligible*year2015*Agency 1961 0.201 1961 0.202 1291 0.017

Eligible*year2015*Collaborators 407 0.561 407 0.562 1208 0.001

Eligible*year2015*Intern 222 0.765 222 0.766 306 0.459

Post*Eligible*year2015*Apprentices

-629 0.497

Post*Eligible*year2015*Standard fixed-term contracts

-1906 0.144

Post*Eligible*year2015*Agency

-1730 0.095

Post*Eligible*year2015*Collaborators

-1796 0.054

Post*Eligible*year2015*Intern

-583 0.551

Large*Eligible*year2015*Apprentices

-25 0.926

Large*Eligible*year2015*Standard fixed-term contracts

3416 0.004

Large*Eligible*year2015*Agency

6065 0.000

Large*Eligible*year2015*Collaborators

2114 0.021

Large*Eligible*year2015*Intern

2586 0.007

Post*Large*Eligible*year2015*Apprentices

72 0.738 Post*Large*Eligible*year2015*Standard fixed-term contracts

-1157 0.413

Post*Large*Eligible*year2015*Agency

-3648 0.016

Post*Large*Eligible*year2015*Collaborators

1044 0.003

Post*Large*Eligible*year2015*Intern

95 0.819 The models also include year and month dummies as well as all interactions between year (2013 and 2014), type of contract, the eligibility status and the size of the firm. The variable post is a dummy equal to 1 for observations from March 2015 onwards.

39