[PDF] The impact of the great recession on the Italian labor market

View - Download The impact of the great recession on the Italian labor market

Previous PDF

Next PDF

The impact of the great recession on the Italian labor market

Francesco D'Amuri

October 28, 2010

Abstract

This article provides an assessment of the e®ects of the great recession on the Italian labor market. The crisis signi¯cantly reduced hirings, while increasing terminations for individuals on °exible work arrangements. Micro-level multiple stochastic imputa- tions, coherent with the likely evolution of the employment rate, show that inequality of gross labor incomes among active individuals will increase. In the short term, this pattern will be partially o®set by highly fragmented income support safety nets avail- able. Labor income inequality is driven by changes in employment, while inequality among the employed seems to be rather unsensitive to overall labor market conditions.

JEL Classi¯cation Codes:J64, J65, J82.

Keywords:Great recession, labor market dynamics, unemployment bene¯ts' system, income distribution. Bank of Italy - Research Department and ISER, University of Essex; Homepage: http://sites.google.com/site/fradamuri/; email: francesco.damuri@gmail.com. I am indebted to I. Faiella for constant advice. I am also grateful to A. Brandolini and A. Rosolia for useful comments and to R. Zizza for providing me the stata codes used to simulate Unemployment Bene¯ts and Wage

Supplementation Fund transfers.Opinions expressed here do not necessarily re°ect those of the Bank of

Italy.

1

1 Introduction

Italy has been, together with Germany, one of the European countries experiencing mildest increases in the o±cial unemployment rate during thegreat recession1, (Arpaia and Curci,

2010). This result could be partly explained by the intense use of government schemes pro-

viding wage subsidies to employees working at reduced hours (Cassa Integrazione Guadagni, CIG), preventing ¯rms from shedding workers to reduce labor input in response to adverse demand shocks, and by the rise of discouragement among job-seekers dropping the labor force, particularly in the South, where participation rates were already lower than in the rest of the country (Cingano et al., 2010). Nevertheless, employment fell by 2.4 per cent in the 2008-2009 interval, the greatest contraction since the 1992 crisis. In Italy, as in the rest of Europe (Arpaia and Curci,

2010) and the US (Elsby et al., 2010), men and young workers were the hardest hit by the

labor market downturn. This article analyzes the labor market °ows and the labor income variations generated by such an unprecedented reduction in labor input. Changes in unemployment rates are usually believed as being determined by both an increase in transitions into and out of employment.

2Recent articles have somewhat questioned

this view. According to both Hall (2005) and Shimer (2007), changes in the probabilities of transitions into employment have had an increasing role in determining unemployment variations: Shimer (2007) ¯nds for example that "movements in the job ¯nding probability account for 75 of °uctuations in the unemployment rate since 1948, rising to 95 per cent during the last two decades".

3Also Elsby et al. (2010) ¯nd that exits from employment

were relatively more modest during the recessions started in 1990 and 2001, but do not interpret this as a secular change in the way recessions impact on labor market °ows. Indeed, they ¯nd that the recession started in 2007, as other severe recessions of the past, has been characterized by a substantial increase in out of employment transitions, driven by layo®s. The analysis of yearly transitions in and out of employment carried out in this paper con¯rms Elsby et al. (2010) ¯ndings for Italy: both into employment and out of employment transitions between the last quarter of 2008 and the same period of 2009. 1

Strauss Kahn (2009) was the ¯rst to refer in this way to the severe economic downturn started with

Lehman Brothers' bankruptcy in September 2008.

2See, among others, Blanchard and Diamond (1990) and Perry (1972).

3The article uses data up to 2007.

2 Yearly job destruction probabilities increased by 0.5 percentage points, going to 7.3 per cent. In relative terms, job ¯nding probabilites had a decrease of similar magnitude (-1.5 points, to 14.8 per cent). Changes in the probability of transitions out of employment varied substantially with the type of contract. The brunt of the adjustment was borne by employees on temporary contracts (a result in line with Arpaia and Curci (2010)): the biggest increases concerned employees on °exible work arrangements including ¯xed term contracts and quasi- employees (+3.5 to 20.5 percentage points), 12 per cent of total employment. The remaining

88 per cent, employees on open ended contracts and self-employed, did not experience major

changes in the probability of having this type of transition. After having identi¯ed the changes in labor market °ows taking place during thegreat recession, the estimated probabilities for transitions in and out of employment are used to simulate, coherently with employment rate dynamics de¯ned ex-ante, the likely evolution of labor income and its distribution in 2010. The analysis focuses exclusively on the labor market, while other sources of income are not taken into consideration.

4According to our

baseline simulations (assuming a 0.6 drop in employment between 2009 and 2010 and con- stant participation rate), gross labor income inequality among active individuals is expected to rise, with a Gini index increasing to 25.4 in 2010 from 24.7 in 2009 as a result of the increase in the number of unemployed, while income distribution among those remaining employed will barely change. We also simulate what would happen to labor income dis- tribution assuming a zero job ¯nding rate during 2010, thus focussing on job destruction and unemployment bene¯ts. In this case, Gini index would increase 3.3 points to 0.281, an increase only partly cushioned by unemployment bene¯ts: accounting for income support schemes, the increase in the Gini index (1.8 points) is less pronounced but still substantial. According to the simulations, standard income support schemes

5have average replacement

rates (varying considerably across sectors and working arrangements) slightly above 30 per cent in the ¯rst quarter of unemployment or suspension of work contract and quickly de- creasing with duration. At the fourth quarter in unemployment, they are as low as 6 per cent, with almost 80 per cent of the displaced completely uncovered by income support 4

For a more general assessment of the impact of the crisis, and of the role of welfare systems in cushioning

income losses for the unemployed, see Figari et al. (2010) and Dolls et al. (2010).

5These include ordinary wage subsidies for workers on reduced hours and unemployment bene¯ts (see

section 3 for a description of the Italian UB system). Discretionary schemes (cassa integrazione in deroga)

and long term unemployment bene¯ts (indennitµa di mobilitµa) not included. 3 schemes. Finally, a stress test of the unemployment bene¯ts system is proposed, assuming di®er- ent employment rates' dynamics obtained modifying the baseline scenario adding +1 and -1 standard deviation of the unemployment rate (1.7 percentage points) for a given par- ticipation rate. One year projections show that average replacement rates, albeit very low and rapidly decreasing over time, do not vary much with the severity of the labor mar- ket downturn. The impact is instead clear when considering income distribution: adding a standard deviation increase in the unemployment rate to the baseline scenario (bringing the unemployment rate in 2010 to 10.9 from 9.1) substantially increases the Gini index of individual gross labor income (+1.7 points to 27.1). This paper is organized as follows: section 2 introduces the dataset employed and data re¯nements, section 3 describes labor market related income support schemes. An analysis of transitions into and out of employment is provided in section 4 while section 5 de¯nes the multiple stochastic imputation method employed for simulating the distributional impacts of di®erent labor market states. Results are provided in section 6, while section 7 concludes.

2 Data

The Italian Labor Force Survey (ILFS), the o±cial dataset used for estimating labor market conditions in Italy, provides the statistical basis for this study. It contains full information on labor market status and other socio-economic characteristics of a large sample representative of the Italian population on a quarterly basis (for a description, see Ceccarelli et al. (2007)). We use data on respondents' labor market status relative to the 2006-2009 period, including the net wage earned by employees (available for year 2009 only). By means of standard mincerian regressions we impute net labor income to self-employed workers as well.

6Using

the longitudinal version of the dataset, we are able to identify with high precision transitions in and out of employment on a 12 month interval. We prefer to impute labor status based on 6 Labor income from self-employment is not reported in the ILFS data. Self employment income is

imputed by means of six weighted mincerian regressions run separately for men and women and for three

geographical areas (North, Centre and South), historically characterized by di®erent labor market conditions.

The dependent variable is the net average monthly wage, while independent variables are: 103 province ¯xed

e®ects, 12 dummies for maximum educational level attained and 7 age class dummies (identifying 10 year

age intervals). 4 the longitudinal dataset rather than to use the recall questions to avoid miss-classi¯cations due to recall error (Bowers and Horvath, 1984; D'Amuri, 2010; Poterba and Summers, 1986). Gross labor incomes are predicted by an auxiliary regression estimated on a tax-form based dataset. 7 In order to correct for miss-reporting, wages are multiplied by a constant factor equalling the estimated total gross wage bill with the corresponding value taken from the national accounts. The same factor is also used for self-employment income, since in this case there is no external source available for validation.

3 The Italian unemployment bene¯t and wage supple-

mentation system Average Labour Market Policy (LMP) expenditure per unemployed as a share of GDP per capita was 11 per cent lower in Italy than in the EU27 average in 2007, according to Eurostat (2009) data (tab. 1). Expenditure was relatively higher (6% more than the EU27 average) in LMP measures (training, job sharing, employment incentives) and early retirement (39% more than the EU27 average), while being lower in public employment services providing advice and information on jobs and training. All the other main west European countries, with the exception of the United Kingdom, spent more than Italy on LMPs: Germany, France and Spain spent respectively 12, 17 and 7 per cent more than the EU27 average. The safety net for individuals experiencing transitions to non-employment is fragmented, lacking a basic framework common to di®erent type of workers and working arrangements. The main non-discretionary schemes for income support are Unemployment Bene¯ts 7

In particular, the auxiliary regressionYLgross=®+¯YLnet+°Y2Lnetis run on an administrative dataset

(MEF, 2009) reporting gross labor income (YLgross) and net labor incomeYLnetearned in 2007 for discrete

intervals of the former. Using the estimates for the parameters®,¯and°, a value ofYLgrossis imputed

at the micro level in the ILFS dataset based on the realizations ofYLnetin the same dataset (the tax code

for did not have any substantial change between ¯scal year 2007 and 2009). The imputation gives a very

good approximation of gross labor income. For employees, according to MEF (2009) data, labour income

constituted 90.5 per cent of total income subject to the Personal Income Tax in the ¯scal year 2007; as a

consequence there is a close link between labour income, total income, and thus tax rates. A more precise

imputation would require a tax-bene¯t model and information on other sources of income, not available in

the ILFS. 5 (ordinary or with reduced requirements), Wage Supplementation (ordinary or extraordinary) and long term unemployment assistance (for further details see Anastasia et al. (2009) and

European-Commission (2009)).

Ordinary unemployment bene¯ts(Indennitµa di disoccupazione) are the main source of support for individuals who have been laid o®, or whose contract expired. Maximum du- ration is 8 months (12 for workers older than 50), while replacement rates, equal to 60 per cent in the ¯rst month of unemployment, are decreasing in the length of the treatment. In order to be eligible the individual need to have paid social security contributions for at least

52 weeks in the last two years. Individuals who do not meet this eligibility criteria, but

who paid social security contributions for at least 78 days in the last two years, can apply for unemployment bene¯ts with reduced requirements, characterized by lower replacement rates. Ordinary Wage Supplementationis a wage subsidy for employees in manufacturing and construction, providing 80 per cent of the wage for a maximum of 13 weeks in which the worker is temporarily not involved in production due to lower demand.

8Actual replacement

rates tend to be lower since the maximum payable amount is equal to 893 euros in 2010 (1073 when gross monthly wage is above 1932 euros). Trainees are excluded. Extraordinary Wage Supplementation provides similar replacement rates for a maximum of 36 months, but is available only for employees working in manufacturing or services ¯rms undergoing closure, bankruptcy or a major restructuring. Eligibility depends on the size of the ¯rm, at least 15 employees in manufacturing, 50 in the services sector. Finally, thelong term unemployment assistance(indennitµa di mobilitµa) is an income support scheme reserved to employees enjoying Open Ended Contracts, having a tenure of at least one year in a ¯rm that has exploited the Extraordinary Wage Supplementation fund, or is undergoing closure or a major restructuring. Duration, increasing with the age of the worker, is up to 3 years, and can be extended for further 12 months when the ¯rm is located in the South of the country (where the economy is weaker). In 2008 the Italian government (Decreto legge of the 29th of November 2008, n.185) has extended the income support schemes to some categories of previously uncovered individuals, introducing three month unemployment bene¯ts with a 60 per cent replacement rate for individuals whose contract is temporarily suspended, provided they have paid social security 8 Extensions up to 12 months can be provided on a discretionary basis. 6 contributions for at least 52 weeks in the previous three years and are not covered by the Ordinary Wage Supplementation scheme. A one-o® payment equal to 20 per cent of last year labor income was also introduced for quasi-employees, a category including formally self-employed individuals actually working as employees mainly for tax reasons and for reducing Employment Protection Legislation. Moreover, Wage Supplementation income support schemes on a discretionary basis (Cassa Integrazione in Deroga) became available for those workers employed in ¯rms not eligible for standard treatments.

4 Transitions in and out of employment

As a starting point we estimate individual probabilities of transition from Employment (E) to Non-employment (NE) and viceversa at one-year intervals. We include in estimation both a pre-crisis interval (2006:1-2008:3) and an interval including the crisis (2008:4-2009:4). For the E to NE (and viceversa) transition equations we estimate the following expression on the full interval: y whereyitfor individualiat timetis missing if the individual was non employed int-4, it is equal to one if the individual is non employed intand 0 otherwise, in the E to NE equation (opposite de¯nition in the NE to E one). The equation is estimated separately for four group of workers: employees with Open Ended Contracts (OEC), employees with

Fixed Term Contracts (FTC), quasi-employees

9and self-employed workers. The dummy

Crisis, equal to one for ¯ve quarters (2008:4-2009:4 interval) and zero for the remaining 13

quarters (2006:1-2008:3), identi¯es the average e®ect of the crisis on transition probabilities

for each of the workers' subgroups. Finally, the matrixXitincludes usual controls for socio- demographic characteristics (in both equations) and job-related characteristics (only in the job termination one). According to our estimates, the crisis signi¯cantly increased the probability of experi- encing a transition out of employment for employees with Fixed Term Contracts and for quasi-employees, respectively by 3.7 and 2.3 per cent (table 2). The impact of the downturn 9

Quasi-employees are de¯ned as formally self-employed workers, actually working as employees for tax

reasons and for reducing EPL. They are de¯ned as self-employed workers working for a single company, in

the company's premises and with a non-°exible time schedule. 7 is not as substantial when we take into consideration self-employed and workers with Open Ended Contracts. Beyond the average e®ects, the impact of the crisis has a di®erent pattern along the 5 quarters of the 2008:4-2009:4 period. The magnitude of the average e®ect of the crisis on transition probabilities needs to be interpreted with caution. Indeed, given that the crisis started in 2008:4, interviewed individuals have been exposed to its e®ects for a di®erent number of quarters depending on the time in which the interview took place (one quarter in 2008:4, two in 2009:1 and so on). Moreover, for the last quarter of the crisis interval (2009:4), labor market status might have been a®ected by the crisis already int¡4, since the downturn started more than one year before. Assuming that those who did not to exit employment (did not ¯nd employment) in the ¯rst quarter of the crisis are a positively (negatively) selected sub-sample of all workers, the estimates of the e®ects of the crisis are not fully comparable with those calculated on an interval starting with a "non-crisis" quarter and ending in a "crisis" one. With these caveats in mind, the ¯rst half of 2009 has been the period in which the transitions out of employment have increased the most for employees with °exible working arrangements. On the other side, workers enjoying open ended contracts did not experience any signi¯cant change in the probability of exiting employment until the the fourth quarter of 2009 (+0.5 percentage points). Turning to transitions into employment, table 3 shows that the probability of ¯nding a job for those not in employment decreased by 1.5 percentage points during the crisis. 10 According to our estimates, during the 5 "crisis" quarters analyzed here, the overall probability of experiencing a transition out of employment has been equal to 5.4 and 5.7 per cent respectively for employees on OEC and for self-employed (tab. 4). This probability is much higher for employees on FTC and quasi-employees (respectively 20.5 and 18.3 per cent). The latter are also the ones experiencing a substantive increase in the likelihood of remaining out of employment due to the crisis (respectively +3.7 and 2.7 percentage points). On the other side, the probability of moving from non-employment to employment falls from

16.2 to 14.8 percentage points (-1.5 percentage points) due to the crisis (tab. 5). These

results underline the fact that changes both in the transitions out and into employment determined the adjustment in employment levels during the crisis, and in particular that variations in °ows out of employment are non negligible. 10

See Shimer (2008) for an analysis of the impact of the business cycle on the probability of ¯nding a job.

8

5 Multiple stochastic imputation and simulations

Based on the speci¯cations of section 4, estimated on the 2008:1-2009:4 interval, we assign to each worker employed in quarterta probability of not being employed in quartert+4 and viceversa, under the assumption that the parameters estimated in equation 1 remain constant during thet,t+4period. Similar LPM models estimate the probability of bene¯ting from the wage-supplementation fund or of experiencing a temporary suspension of the work contract in quartertfor individuals employed in quartert(for a de¯nition of the relevant income support schemes see section 3). After active individuals are assigned a probability for each of these four events, we simulate the evolution of main labor market variables during thet,t+4period for four di®erent scenarios, assuming a constant participation rate and di®erent patterns for the employment rate. This include a baseline scenario in which employment decreases 0.6 per cent, one with constant employment, two more scenarios based on the baseline one plus (minus) one standard deviation in the unemployment rate (1.7 points).

11Simulations are

based on the estimated probabilities for each of the four events and amultiple stochastic imputation(Rubin, 1996).12 In the ¯rst simulation, it is assumed that the employment rate remains constant in each of the four quarters of thet,t+4period. Total gross labor income, its distribution, and unemployment bene¯ts paid might still change: even for a constant employment rate, individuals are still experiencing transitions into and out of employment, and their labor income and unemployment bene¯t entitlements might be di®erent. Moreover, there is a continuum of probabilities of experiencing the Employment-Non Employment and Non- Employment to Employment transitions for a given employment rate. We assume that these probabilities remain unchanged as estimated on the(t,t-4)interval in equation 1, from which we numerically derive quarterly probabilities of transition that match the 4 scenarios for the employment rate for the period(t,t+4). In order to impute labor market state in each of the 4t,t+4quarters, a random draw is made from a uniform distribution [0;1]. If the realization of the draw is lower than the quarterly probability of experiencing the event, based on the estimate of equation 1 and the numerical approximation, the individual changes its state. For example, suppose a worker has been assigned a 0.1 probability of 11 Standard deviation is calculated on the 1993-2009 period.

12See the appendix for a brief description of the imputation procedure.

9 an employment to non-employment transition betweentandt+ 1, given she is employed int. If the random draw gives a value smaller than 0.1, the individual becomes non- employed. In the same quarter, this process is repeated for Non Employment to Employment transitions (conditional on being non employed in previous quarter), and for being on Wage Supplementation or Suspension (conditional on being employed in the same quarter). Conditional on imputed labor market status int+ 1, this exercise is repeated for tran- sitions occurring betweent+ 1 andt+ 2 and so on, untilt+ 4. It is important to recover the labor market conditions in each of the quarters of the simulation, and not only at yearly intervals, since unemployment bene¯ts and transfers from the wage supplementation fund crucially depend upon the duration of the event at study. In the simulation we assume 100 per cent take up, while eligibility criteria for these transfers follow those described in section 3. Moreover, in order to assess the robustness of the results, we repeat the simulation 30 times with 30 di®erent randomly generated draws from the uniform distribution. This multi- ple imputation exercise assesses the variability of the imputation; a more general assessment of total variance should also include the variability due to the sampling nature of the esti- mators, but this would be out of the scope of this paper. In ¯g. 1 we show the distribution of the estimates for the overall employment level int+4resulting from the multiple im- putation over thet,t+4period. Estimates, obtained assuming constant employment, show some variability, clustering at a value slightly below 23 million. Mean and median of the 30 estimates are quite close (22924 and 22922 thousand of employed people), while the 10th and 90 the percentile are in an interval of -0.2 and +0.2 per cent with respect to the mean. In the other three simulations, we assume di®erent scenarios for evolution of the employ- ment rate. The starting point is common for all the simulations, and it is equal to actual labor market conditions for 2009:4 (employment and unemployment rates respectively equal to 57.1 and 8.6 per cent). In the baseline simulation, the number of employed people drops by 250 thousand (-0.6 per cent) in the 2009:4-2010:4 period, with the unemployment rate reaching 9.1 per cent in the last quarter of 2010, assuming activity rates remain unchanged in the whole period. The two remaining scenarios are simulated assuming the unemployment rate is higher (lower) than the baseline scenario by one standard deviation (1.7 percentage points). Again, we derive numerically the quarterly NE to E and E to NE probabilities coherent with these labor market developments. In all simulations, a 2 per cent annual 10 growth of labor income for those who are employed is assumed.

6 Results

6.1 Aggregate projections

According to the simulation, when the employment rate remains constant, the increase in total gross labor income is equal to 2 per cent, in line with the nominal labor income increases assumed (table 6). This means that, when the employment level is constant, the recomposition of workers through °ows in and out of employment leaves unchanged average labor income. In order to focus on the e®ects of job destruction (and of the unemployment bene¯ts system), we also project labor income assuming no transitions from non employment to employment. In this case, total gross labour income decreases by 1.3 per cent. When including Unemployment Bene¯ts and Wage Subsidies for workers on reduced hours, the dynamic of this aggregate turns slightly positive (+0.1 per cent). The picture is di®erent when assuming a 0.6 per cent drop in employment in the 2009:4-

2010:4 period. Under the hypothesis that participation rate remains constant, this would

entail a 0.6 percentage points increase in the unemployment rate at 9.1 per cent. In this case, total gross labor income is expected to rise by 1.5 per cent, thanks to the contribution of the 2 per cent annual nominal growth assumed. This variation would be equal to -0.4 and +3.0 per cent respectively when considering a scenario with plus/minus one standard deviation in the unemployment rate (table 7). Going back to the baseline scenario, when conditioning on individuals employed in 2009, the drop in labor income is more pronounced (-1.9 per cent), cushioned when including Unemployment Bene¯ts and Wage Subsidies (-0.4 per cent). Actual coverage of these income stabilizers would be higher, since some of the schemes (indennitµa di mobilitµa and cassa integrazione in deroga) are not included in the simulation. Finally, it is reassuring to note that results are very similar when considering alternatively the mean or the median of the 30 estimates obtained through the multiple stochastic imputation. 11

6.2 Unemployment bene¯ts

In the baseline scenario, average replacement rates are slightly above 30 per cent in the ¯rst quarter of unemployment/suspension (table 8). Wide variation emerges across sectors and contract types (see ¯g. 2), with coverage ranging from 0 to 80 per cent. This is a well known feature of the Italian income support system (Anastasia et al., 2009). Coverage quickly drops with time in unemployment or suspension: after four quarters of continuous unemployment/suspension, average replacement rates are around 6 per cent, with almost

80 per cent of the displaced receiving no bene¯ts at all.

The stress testing of the unemployment bene¯ts gives interesting results: average re- placement rates, and the percentage of displaced workers with no coverage does not change much with labor market conditions in the four simulations assuming the same level of un- employment rate in 2009:4 (8.6 per cent) and rates between 7.6 and 11 per cent in 2010:4. Albeit very low, average replacement rates in the ¯rst month of unemployment remain around 33 per cent irrespective of the unemployment rate. The percentage of uncovered displaced workers is slightly more sensitive to changes in the labor market conditions, being

16 per cent when assuming 7.6 per cent unemployment rate in 2010:4 and 4 points higher

when unemployment is at 11 per cent.

6.3 Labor income distribution

Going back to the baseline scenario, the e®ects of the crisis will be felt among workers earning less than 3000 euros per month (approximately the lowest 80 per cent of the labor income distribution among the labor force, see ¯g. 3). Once UB and WS transfers are included, income losses are cushioned, but there is still a considerable amount of displaced individuals with gross income lower than 1000 euros per month, or no income altogether (¯g. 4 and 5). Always in the baseline scenario, the crisis will increase labor income inequality, with a Gini index calculated on active (employed plus unemployed) individuals raising from 24.7 in 2009:4 to 25.4 in 2010:4 (tab. 9). Following Atkinson and Brandolini (2006), we can decompose the increase in inequality in a component due to the rise in unemployed indi- viduals, and a component due to the change in inequality among those who remain in the labor force:GA=u+ (1¡u)GE, whereuis the unemployment rate, whileGAandGEare 12 the Gini indexes of gross labor income, calculated respectively among active and employed individuals. It can be noted that all the increase in inequality is due to the increase in unemployment, while wage inequality among those who stay employed barely changes. In order to focus on the impact of job destruction on inequality, and the role played by unemployment bene¯ts, we also simulate the Gini index setting to zero NE to E transitions. In this case the the value of gross income inequality in 2010 would be substantially higher at 0.281 (3.3 points higher than in 2009). Including UB and WS transfers, the increase in labor income inequality is less pronounced but still substantial (1.8 points to 0.266). Further interesting results are obtained through the other simulations. In particular, it is worth mentioning the fact that, even when unemployment rate is assumed to increase from 8.6 to 11.1 in one year (implying substantial changes in the composition of employed people) gross labor income inequality barely changes conditional on the individual being employed. All the increase in inequality is due to the increase in unemployed individuals.

7 Conclusions

In the 2008:4-2009:4 period, thegreat recessionhad an impact both on job ¯nding and job termination probabilities. Transitions out of employment increased in particular among employees with °exible contracts. The fragmented Italian unemployment bene¯t system is able to cushion wage losses only in the short run and for selected categories of workers. Nevertheless, a stress test shows that labor market related income stabilizers' replacement rates and coverage, albeit low and quickly decreasing over time, do not vary much with overall labor market conditions. Finally, according to the simulations, the increase in labor income inequality due to the crisis is accounted for by the drop in employment, while inequality among those remained em- ployed does not vary much. This means that the change in the composition of employment, entailing a reduction of the number of workers with °exible work contracts, does not change much inequality among active individuals, that is instead driven by the overall employment level. 13

References

Anastasia, B., M. Mancini, and U. Trivellato (2009). Il sostegno del reddito dei disoccupati: note sullo stato dell'arte. Tra riformismo strisciante, inerzie dell'impianto categoriale e incerti orizzonti di °exicurity.ISAE Working Paper, 112.quotesdbs_dbs22.pdfusesText_28

[PDF] unemployment rate in africa 2019

[PDF] unemployment rate in india

[PDF] unemployment rate problem solution essay

[PDF] unemployment rates in great recession

[PDF] unesco cse guidelines 2018

[PDF] unfair treatment of employees in the workplace causes all the following except ________.

[PDF] unfccc cdm

[PDF] unfccc cdm forms

[PDF] unfccc cop25 outcomes

[PDF] unfccc definition of climate change pdf

[PDF] unfccc initiatives

[PDF] unfpa cse

[PDF] unfpa full form

[PDF] unfpa logo

[PDF] unfpa objectives