How Shortening Potential Benefit Durations Affects ... - SFI

sfi.dk

How Shortening Potential Benefit Durations Affects ... - SFI

How Shortening Potential Benefit Durations Affects

Unemployment Spells: Evidence from a Natural

Experiment

Jan C. van Ours ∗ Milan Vodopivec †‡

August 11, 2005

∗ Tilburg University, CentER, IZA, and CEPR; corresponding author:

Department of Economics,

Tilburg University, P.O. Box 90153, 5000 LE Tilburg, The Netherlands; email: vanours@uvt.nl

† World Bank; email: Mvodopivec@worldbank.org

‡ The authors wish to thank The National Employment Office of Slovenia and the Statistical Office of

Slovenia for providing the data used in this study. Jakob Tomse provided excellent assistance in organizing

the data. Support from the World Bank research project “Incentive Effects of UI systems in Transition

Countries” (RF-P087059-RESE-BBRSB) is gratefully acknowledged. The authors are grateful to participants

in seminars held in Canberra (RSSS), University of Melbourne, University of Sydney, Brussels (EU),

Tilburg, the IZA prize ceremony (Berlin), Maastricht, and Florence (EUI) for their comments on a previous

version of this paper (Van Ours and Vodopivec, 2004), as well as to participants of seminars in Paris

I (TEAM), The Hague (CPB), and Zurich for comments on a more recent version.

1


Abstract

The authors of this paper investigate the disincentive effects of shortening the potential

duration of unemployment insurance (UI) benefits. We identify these disincentive

effects by exploiting changes in Slovenia’s unemployment insurance system – a

“natural experiment” that involved substantial reductions in the potential duration

of benefits for four groups of workers plus no change in benefits for another group

(which served as a natural control). We find that the change had a positive effect on

the exit rate from unemployment – to new jobs and other options – for unemployment

spells of various lengths and for several categories of unemployed workers.

Keywords: Unemployment insurance, potential benefit duration, job-finding rates

JEL-codes: C41, H55, J64, J65

2


1 Introduction

In theory, unemployment insurance provides a disincentive for job-seeking, which is affected

both by the level of benefits (relative to the expected wage) and by the potential benefit

duration (PBD). The greater the level of benefits, the less costly the period of the job

search, so workers tend to search for jobs less intensely and tend to remain unemployed

longer. Putting a limit on the duration of benefits tends to speed up the job search. As

the date approaches when benefits will expire, unemployed workers tend to increase the

intensity of their job search and the rate of job-finding increases. Moreover, the exhaustion

of benefits creates a ”spike” in the exit rate from unemployment.

Basing his work on a job search model and assuming that non-market time and market

goods used in household production are substitutes for household production, Mortensen

(1977) provides a theoretical explanation for this phenomenon. 1 Other suggested explanations

include the strategic timing of job starting dates and implicit contracts between

unemployed workers and their previous employers, in which the employers rehire the workers

at about the time their benefits expire (Card and Levine, 2000).

In line with theoretical predictions, empirical studies confirm the link between the PBD

and the duration of covered unemployment. 2 Katz and Meyer (1990), for example, estimate

that a one-week increase in PBD increases the average duration of unemployment for U.S.

recipients of UI by about one day. Card and Levine (2000) report a disincentive effect of

about 0.5 day per additional week of PBD, also based on U.S. data. Lalive and Zweimüller

(2004) report a disincentive effect of about 0.4 day for Austrian benefit recipients. The

PBD affects not only the duration of unemployment but also the pattern of the exit rate.

Many studies find a sharp increase in the exit rate from unemployment just before benefits

expire. 3

1 If non-market time and market goods are complements there is an upward shift.

2 See Atkinson and Micklewright (1991) for an overview of theoretical and empirical evidence and Lalive

et al. (2004) for a recent overview of empirical studies.

3 Katz and Meyer (1990), Card and Levine (2000), and Addison and Portugal (2004) find such “spikes”

for U.S. benefit recipients. Carling, Edin, Harkman and Holmlund (1996) find spikes for Sweden in both

the job-finding rate and the exit rate from unemployment to labor market programs. Roed and Zhang

3


Although the empirical literature has demonstrated an association between the potential

and actual duration of UI benefits, it has mostly failed to plausibly identify PBD’s

causal relationship with exits from unemployment. Probably the strongest evidence comes

from studies showing spikes in job-finding rates just before benefits are exhausted, which

strongly suggests a causal effect. But according to Card and Levine (2000), those spikes

existed at the same points even when the PBD changed. Many researchers take advantage

of variation in maximum durations triggered by the U.S. business cycle, but the same forces

responsible for a benefit extension could also reduce the exit rate from unemployment. To

our knowledge, the only exception in the U.S. literature is Card and Levine (2000), who use

a politically motivated extension of unemployment benefits in New Jersey – an extension

unrelated to labor market conditions. As in previous literature, they also find that longer

maximum benefit durations increased the duration of a spell of unemployment, but the

established effect was smaller.

Our paper takes advantage of another credible, independent source of variation in PBD:

a change in Slovenia’s UI benefit law. Faced by increasing unemployment, among both

UI recipients and the long-term unemployed, Slovenia in October 1998 drastically reduced

the potential duration of benefits, roughly by half for most groups of recipients. Before

reform, for example, workers with 5 to 10 years of work experience were eligible for benefits

for up to 9 months, and workers with 10 to 15 years of experience were eligible for up to

12 months of benefits. After the change, both groups of workers were eligible for only up

to 6 months of benefits. Similar to Card and Levine (2000), our analysis builds on the

quasi-experiment as a basis for identifying treatment and control groups. Not only did

the reform introduce a variation in PBD unrelated to the state of the labor market, but it

also changed PBD differently for different groups of beneficiaries. The PBD was reduced

dramatically for most groups of workers but remained unchanged for workers with less than

a year and a half of work experience, creating a natural group through which to control

for the effect of changes in the labor market and other conditions, when comparing exits

(2003) find end-of-benefit spikes for Norway, and Lalive and Zweimüller (2004) and Lalive, Van Ours, and

Zweimüller (2004) for Austria.

4


from unemployment before and after the law changed.

Based on this “natural experiment” with PBD, we identify a significant increase in

the exit rate out of unemployment – both to employment and to other destinations – at

various durations of unemployment spells. We also identify a clear spike in the exit rate

from unemployment in the month unemployment benefits expire. These results are in line

with many others in the literature. But in addition to providing new evidence about the

changes in survival and hazard rates associated with changes in PBD, arguably the major

contribution of this paper is to clearly pinpoint causality – namely, that shortening the

PBD caused the exit rates to increase.

Causality is investigated formally by econometric testing, but nearly as forceful is a mere

comparison of the pre- and post-reform survival functions of benefit recipients. Figure 1

depicts these functions for several groups of “otherwise identical” individuals (grouped by

work experience), before and after reform. As Figure 1 shows, post-reform survival rates

improved slightly for the least experienced workers (the control group), for whom the PBD

has not changed. Other groups (which we refer to as treatment groups, because the reform

shortened their PBD) saw greater improvements in survival rates, with the biggest changes

occurring for the groups whose maximum benefit duration changed the most. The control

group’s improved hazard rate can be attributed to improved labor market conditions and

other policy changes after reform, but for this explanation to hold for the other groups

the changes in underlying labor market conditions would not only have to shift the spikes

differently for each group, but to shift just “the right way.” Why would labor market

conditions improve more for more experienced workers? And why would the precipitous

drops in UI survival function (corresponding to spikes in hazard rates) all shift from the

point where benefits used to run out before reform to the point where they run out after

reform? That this particular pattern of shifting occurs for all treatment groups virtually

rules out alternative explanations.

Moreover, although the exit rate from unemployment for the control group (the youngest

and least experienced group) improved after reform, exit rates improved even more for the

treatment groups (comprising older, more experienced workers). If changing labor market

5


conditions were the reason for improved exit rates, relative performance of younger workers

who are particularly sensitive to business cycles should have been better after reform than

for other groups, yet our findings suggest the opposite.

In the next section of the paper we describe the 1998 change in Slovenia’s unemployment

insurance system. In section 3 we present our data and explanatory variables and in section

4 the results of our analysis of survival functions. In section 5 we discuss the results of our

analysis using hazard rate models to identify the effects of PBD. We present our conclusions

in section 6.

2 The changes in Slovenia’s UI system

Formerly a part of Yugoslavia, Slovenia – a country with a population of only 2 million

– became independent in 1991 and joined the European Union in 2004. Since 1995, its

unemployment rate has remained remarkably stable, at a level of 6 to 7%. Like OECD

countries, Slovenia provides income support for the unemployed through a social insurance

program that combines unemployment insurance (UI) and unemployment assistance (UA).

Benefits cover the majority of employed persons, whatever their industry or occupation (the

most notable exception being the self-employed). Benefits under employment insurance

are earnings related, with the duration of entitlement contingent on work experience, with

predetermined maximum and minimum levels. Benefits under UA are means-tested. They

are offered to those who have exhausted their eligibility for UI (but are still unemployed)

and to selected groups of other workers who do not qualify for unemployment insurance

benefits. Benefits are financed mostly by the budget, with token contributions paid by

employers and workers.

Registering as unemployed with an employment office entitles workers not only to cash

benefits but also to other services: information about and referrals to job vacancies; vocational

counseling and assistance with social and psychological adaptation; reimbursement

for eligible travel and medical check-up expenses associated with job search; and the opportunity

to participate in a host of active labor market programs. Active labor market

6


programs include training courses, job-search assistance, job clubs, self-employment programs

(offering free advisory programs and financial assistance), and hiring subsidies (some

available only to the long-term unemployed). One obvious reason to remain registered after

unemployment insurance benefits run out is to collect unemployment assistance, but there

are many reasons why unemployed workers tend to remain registered for the full period of

unemployment.

As the number of unemployed, UI recipients, and long-term unemployed kept rising,

Slovenia in October 1998 reformed its unemployment benefit system. Arguably the most

significant change was to reduce the potential duration of benefits. Under the new system,

the length of the UI entitlement period was shortened roughly by half for most groups of

recipients. Before the reform, for example, workers with 5 to 10 years of work experience

were eligible for 9 months of benefits, and workers with 10 to 15 years of experience were

eligible for 12 months. After reform, both groups were eligible only for 6 months of benefits.

A notable feature of reform was that different groups of beneficiaries were treated differently

– a trait we take advantage of in testing reform’s effects.

Reform also called for several measures aimed at speeding up benefit recipients’ reemployment,

including improvement in employment services, the obligatory preparation of a

re-employment plan for each benefit recipient, and more frequent contact between counselors

and recipients. Furthermore, reform broadened the definition of a suitable job (after

4 months, the unemployed could be offered worse-paying jobs or jobs requiring a substantial

commute) and introduced stiffer sanctions for refusing job offers. It also called for

stricter monitoring of eligibility. Benefit recipients had to make themselves available to

employment office counselors several hours a day. For the first time, inspectors (a special

arm of employment offices) would check to see if benefit recipients were in fact unemployed

(among other things, by paying home visits to benefit recipients) and actively searching

for a job.

Reform restricted access to UI benefits at the same time that it made participation

in active labor market programs easier and more attractive. Public works participants

were given the status of regular workers giving them access to many fringe benefits (such

7


as vacation and pension coverage). A hiring program that reimbursed employers for the

payment of social security contributions was strengthened by broadening the target groups

(to include the long-term unemployed, first-time job-seekers, older workers, and recipients

of unemployment benefits) and increasing the amount of reimbursement. And after the

law was changed, the government spent more on active labor market policies. Spending

on those policies as a share of GDP increased from 0.40 percent in 1998 to 0.52 percent in

1999.

3 Data, comparison groups, and variables

3.1 A dataset based on administrative records

The empirical analysis is based on the administrative records of unemployment spells,

combined with selected information on formal employment spells. Included are all spells

of unemployment benefit recipients that started between August 1, 1997, and December

31, 1999 (with censoring on December 31, 2001). For each spell, the database contains

the starting and end date for registered unemployment. The destination of exit (to a job

or something else), and information about the receipt of unemployment insurance benefits

(the starting and end date for eligibility and the actual end date for receipt of benefits).

Personal and family characteristics of benefit recipients are also included. To improve

information about the end of unemployment spells, we used a separate source: a work

history data set for formal sector workers.

Our database is unusually rich in high-quality information. For one thing, it provides

complete coverage: all workers registered as unemployed for the selected period are included.

For the analysis, we selected a random sample of about 6 percent of spells of

unemployment. For another thing, being administrative in nature, the information is free

of problems common in survey data (such as non-response and interviewer bias). Finally,

accurate information at our disposal covers not only the period when workers were covered

by unemployment benefits but also the period of transition from unemployment to

8


employment after benefits expired. Unlike many studies in which information about the

job-finding date is based on unreliable self-reporting of the unemployed workers themselves

(they have little incentive to report accurately), we have independent information

from employers about when the post-unemployment period started.

After removing from the database individuals for whom there is incomplete information,

we have information about 9,196 males and 10,853 females (see Van Ours and Vodopivec,

2004, for details). We distinguish between job finding and other exit destinations. The

destination “to job” is defined by administrative records as exit to private or public employment,

including self-employment, with or without the mediation of employment offices.

Other destinations include exits to active labor market programs (public work programs,

subsidy programs), voluntary and involuntary removal from registration (voluntary withdrawal,

refusal to report, turning down a suitable job, not actively seeking employment,

unavailable for work) and a miscellaneous category that includes maternity leave, military

service, and death. On average half of the non-job exits involve labor market programs

and about one-third involve removals from registration.

Our overview of unemployment dynamics for the selected sample (table 1) distinguishes

between the cumulative probability of outflow to a job, outflow to other destinations, and

total outflow, after 3, 6, 9, and 12 months of unemployment, before and after the change in

the unemployment benefits law. For example, the cumulative probability of having found

a job within 6 months was 45.8% for males before the law changed and 51.0% after. The

cumulative probability of exiting from unemployment after 6 months was 3.3% before the

law changed and 12.0% after. For both men and women, however long they qualified for

benefits, increasing percentages exited from unemployment. On overage, the proportion

who exited to active labor market programs increased from about 42 percent before the

law changed to 58 percent after, and the proportion removed from registration fell from 39

percent to 28 percent.

9


3.2 How comparison groups were formed

Our data show that the probability of exiting from unemployment increased after reform

of the UI law, but how could we disentangle the effects of changing the potential duration

of benefits from the effects generated by changes in labor market conditions (such as a

general increase in demand for labor), other policy changes, or a the combination of these

factors? We adhere to the ”difference in differences” approach: We compared outcomes

before and after reform, exploiting the exogenous variation in potential benefit durations

introduced by the reform, and comparing outcomes for a natural control group to outcomes

for several treatment groups. In short, we exploited specific design features of the changes

in UI law. 4

Our basic strategy was to compare benefit recipients’ exit rate from unemployment

before and after the law changed. The legislative change provides an exogenous variation

in PBD. To reduce a possible selection bias in forming “before and after” comparison

groups, we examined whether expectations of the law’s introduction affected inflows from

employment to unemployment. And our analysis indeed showed that the contemplated

law reducing the potential duration of UI increased inflow into unemployment just before

the new UI law was introduced, and reduced it right after (see for details Van Ours and

Vodopivec, 2004). So, to avoid biased estimates we did not consider data for the period

of two months before and two months after the law’s introduction. We chose an inflow

sample for the period August 1, 1997 – July 31, 1998 as our “before group,” and an inflow

sample for the period January 1, 1999 – December 31, 1999 as our “after group” (all spells

are censored on December 31, 2001). Both samples cover a year of inflow, which takes care

of seasonal differences in the composition of the inflow.

To be able to unambiguously identify causality underlying improved exit probabilities,

4 Note that another feature of the Slovenian UI system – that PBD vary across groups of workers –

does not identify unambiguously the source of differences in exit rates. According to the law, individuals

with longer work experience are entitled to longer PBD. But individuals with a longer work history are

typically older, so the fact that they find jobs at a slower rate could be attributed not only to their longer

PBD but also to their being older.

10


we cannot rely solely on the exogenous variation of PBD introduced by the change in UI

law. Even if we observe, for example, that shortened PBD is associated with improved exit

from unemployment, we cannot rule out the effects of changes in underlying conditions in

the labor market (perhaps as a consequence of the business cycle). Fortunately, in our case

the legislative change introduced different rules for different groups of unemployed workers,

and left the PBD of one group (workers employed less than a year and a half) unchanged.

The legislative design allowed us to form one control group (members for whom the PBD

has not changed) and four treatment groups (whose members were subject to different

PBD before and after the law changed). We assume that the relative effect of changes in

labor market conditions on the job-finding rate is the same or similar for all categories of

workers.

In general, one treatment group suffices, but the natural presence of four treatment

groups makes our conclusions more convincing. One could object, claiming that changes

in underlying labor market conditions could affect a control and a single treatment group

differently – in our case, that such changes helped older workers more than younger, and

that such changes just happened to generate a “just right” improvement for the treatment

group (making the spike in the hazard rate coincide with the exhaustion of benefits after

the reform). But such an objection becomes untenable once there are multiple treatment

groups with different changes in PBD, all experiencing similar, specific shifts in their hazard

functions.

So our empirical analysis rests on data about five groups of unemployed workers: one

control group and four treatment groups. Within each group, some unemployed started

to collect benefits before the law changed, and some after. Had the law not changed, all

members of a group would be entitled to the same PBD. Because some of the recipients

in a group registered after the law changed, their entitlement was much reduced in duration.

The exception is the control group (all workers with less than 18 months of work

experience), for whom the PBD stayed unchanged after the law changed.

Table 2 shows the ‘old’ and ‘new’ benefit entitlements for the control group and four

11


treatment groups. The five groups differ in previous work experience, age, or both. 5 For

the control group the potential benefit duration stayed unchanged at 3 months. For the

first treatment group, which work experience of 1.5 to 5 years, the PBD was reduced from

6 to 3 months. The maximum benefit duration was similarly reduced for the other three

treatment groups. Implicitly, as Table 2 shows, experience is strongly correlated with age.

The older workers are, the more work experience they have, and the longer their potential

entitlement to unemployment benefits.

Table 2 also shows how reform affected the duration of unemployment. After reform,

the median duration of unemployment is shorter than before – the size of the reduction in

length of unemployment reflecting the size of the reduction in PBD. Whereas the median

period of unemployment of the control group fell only with 0.7 months, after reform the

median period of unemployment for the fifth (most experienced) group fell with 5.7 months,

from 12.1 months before reform to 6.4 months after.

3.3 Definition of variables

Data on males and females were analyzed separately to account for possible differences in

labor market behavior. Similarly, the following personal characteristics were taken into

account:

• Age: continuous variable.

• Education: dummy variables, Education2 = elementary school, Education3 = vocational

school, Education4 = high school or more, reference group = elementary school not finished.

• Family situation: dummy variables, Family1 = 1 dependent family member, Family2 =

more than 1 dependent family member, reference group = no dependent family members.

• Ill health: dummy variable derived from information employment office counselors got from

interviews with benefit recipients.

5 This paper uses information about workers with working experience up to 20 years. There are 5 such

groups. For information about groups with more work experience see Van Ours and Vodopivec (2004).

12


• Previous working experience: represented by dummy variables for eligibility groups: 1.5–5

years, 5–10 years, 10–15 years, 15–20 years, reference group = less than 1.5 years.

• Shift in PBD: dummy variables: from 6 to 3 months, from 9 to 6 months, from 12 to 6

months, from 18 to 9 months; reference group = PBD remained unchanged.

We expected these personal characteristics to affect the exit rates in various ways, as

discussed in more detail below. We also investigated the extent to which a change in

business cycle/labor market conditions affected exit rates from unemployment. For this

we use a dummy variable related to the year of inflow into unemployment; i.e. dummy

variables for 1999 where the reference group refers to individuals becoming unemployed in

the period July 1, 1997 – June 30, 1998. We refer to this variable as “after the change of

law.” Our identifying assumption is that the effect of conditions in the labor market is the

same for every worker, especially for every entitlement group. The coefficient of the “after

the change of law” dummy variable is identified through the control group for which the

entitlement period did not change.

4 Exploratory analysis – Survival functions

4.1 Stylized facts

After Slovenia’s unemployment insurance law changed, the unemployed tended to leave

unemployment more quickly, as figure 1 shows. For each of the five groups (the control

group and the four treatment groups) there is a separate graph representing the survival

probabilities before and after the UI law changed. For all groups the survival probabilities

after the law changed are smaller than before.

Several features are worth pointing out. Figure 1a, showing the survival functions for

the control group (for which the PBD did not change), illustrates how changes in labor

market conditions affect the job-finding rate. Here the two lines are not far apart, suggesting

that changing conditions in the labor market had only a small effect on unemployment

exit rates. For all the other groups there is a substantial difference between the two lines,

13


which suggests that shortening the potential benefit period stimulated exits from unemployment.

Another obvious pattern is the positive relationship between potential benefit

duration and survival probability. Groups for whom the potential duration of benefits was

long were more likely to remain unemployed. More important, for many groups there is

a substantial drop in the survival probability in the month when benefits expire. Finally,

benefits expire for a substantial fraction of the unemployed workers while they are still

unemployed. For the youngest group of workers, for example, about 60 to 65% are still

unemployed when their benefits expire. For the oldest group in our sample, only about

35% are still unemployed when their benefits expire.

4.2 Exploratory analysis

To get an idea of the effects of reducing PBD without imposing too much structure in the

empirical model, we performed a number of logit analyses on the probability of leaving

unemployment within a particular time period.

Pr(t < t r ) =

ex′ γ

1+e x′ γ

Pr(t ≥ t r ) = 1

1+e x′ γ

(1)

where t refers to the completed duration of unemployment, t r to a threshold (3, 6, 9, or

12 months), x is a vector of explanatory variables, and γ a vector of parameters. We

estimated the parameters using the method of Maximum Likelihood (see Table 3).

We discuss the results only for males because the effects are by and large the same for

females. As Table 3 shows, the state of the labor market (“after the law”) has a positive

effect on the probability of workers leaving unemployment. Age, education, and having

dependent family members have negative effects, which are not significantly negative for

each of the outflow probabilities, but ill health does have a significant negative effect for

them all. Experience has a non-linear negative effect on the outflow probability, although,

again, it is not always significantly different from zero.

The main variables of interests are those that indicate that the effects of reducing the

potential duration of benefits are positive but not always significantly different from zero.

The probability of leaving unemployment within 3 months is affected only by reducing

14


PBD from 6 to 3 months, and not by other reductions. Reducing benefits to 3 months

moves the spike in the exit rate, which apparently causes the overall effect to be significant.

Similarly, reducing the PBD to 6 months has a significant effect on the probability of leaving

unemployment within 6 months, and reducing the PBD to 9 months has a significant effect

on the probability of leaving unemployment within 9 months. If an unemployed worker

receives a benefit under the old system but not under the new, there may be compensating

effects. In this period after the law changes there is an incentive to find a job quickly

because benefits have expired, before the law changes there may be an incentive to search

harder because benefits have almost expired. Logit estimates suggest that reducing PBD

increased the outflow from unemployment, and to understand the mechanisms involved we

analyze a more detailed model of unemployment dynamics. This we discuss in the next

section.

5 Models of hazard rates

5.1 Stylized facts

To understand the mechanisms through which reducing PBD affects unemployment dynamics,

we analyze hazard rates – that is exit rates from unemployment. Analysis of

monthly job-finding rates before and after the UI law changed (figure 2) show interesting

variations among the control and treatment groups. For the control group there is a clear

spike in the exit rate out of unemployment after three months, when benefits expire. We

also observe a spike at 3 months for the treatment groups, as well as spikes when benefits

are exhausted. For the first treatment group there are two spikes in the exit rate: one at

3 months, when the unemployment benefit replacement rate drops from 70% to 60% and

one at 6 months, when unemployment benefits expire. For the other treatment groups,

there is a spike at 3 months, and two others that coincide with the exhaustion of benefits –

one before and one after the change in the law. Exit rates to other destinations before and

after the law changed are small but here too there are end-of-benefits spikes (see figure 3).

15


And the ‘after’ exit rates are higher than the ‘before’ exit rates. This mainly reflects an

increase in the exit rates to active labor market policy programs.

5.2 Modeling changes in the potential duration of benefits

Shortening the duration of unemployment insurance benefits clearly affects the job-finding

rate. The value of unemployment declines as the worker approaches the date when benefits

ends, and the job-finding rate increases because of increasing search intensity and decreasing

reservation wages (Mortensen, 1977).

In a steady-state environment, after benefits

are exhausted the job-finding rate stays constant at a higher level than before. Reducing

the PBD has two effects. First, because the value of unemployment is lower than before,

the initial job-finding rate is somewhat higher than before. Second, because benefits are

exhausted earlier, the job-finding rate reaches a higher level earlier. 6 Figure 4 gives a stylized

representation of these two effects, comparing the job-finding rate of two unemployed

workers A (After) and B (Before) who differ only in terms of their PBD (T A < T B ). For

t ∈ (0,T A ), θ A is somewhat higher than θ B , an effect which we define as the pre-expiration

effect of a shorter PBD, indicated with δ 1 . For t ∈ (T A ,T B ) worker A is likely to have a

higher job-finding rate because his benefits have expired and he has a stronger incentive

to find a job than worker B, who is still entitled to benefits. This is what we call the postexpiration

effect of the reduction in PBD, indicated with δ 2 . Beyond T B the job-finding

rate of both workers is likely to be the same. 7

Because δ 2 represents the effect of having a benefit compared with not having a benefit,

while δ 1 represents the effect of having a short-lasting benefit compared with having a

longer lasting one, δ 2 is likely to be larger than δ 1 but not necessarily. The relative size

of the two effects is an empirical matter. Using hazard rate models allows us to identify

both effects (see Van den Berg, 2001) for a recent overview of hazard rate models). The

job-finding rate at unemployment duration t conditional on observed variables x is assumed

6 And of course in addition to that the end-of-benefit spike occurs earlier.

7 For the moment, we ignore a potential difference in entitlement effect where worker B has a higher

job-finding rate because his future value of unemployment is higher because of the longer PBD.

16


to have the following specification

θ e (t | x) = λ(t) exp(x ′ β) (2)

where β is a vector of parameters and λ represents individual duration dependence, which

is modeled in a flexible way by using step functions:

λ(t) = exp(Σ k µ k I k (t) + δ 1j I j (t) + δ 2 I p (t) + δ 3 I s (t)) (3)

where k (= 1,..,N) is a subscript for duration interval and I j = (1,2,3) is an indicator of the

length of the period after the PBD’s reduction (j=3,6,9), I p is an indicator of the duration

period after benefit expiration, and I s is an indicator for the month of benefit expiration

(s = 3, 6, 9, 12, 18). The µ parameters measure the pattern of duration dependence and

the δ parameters indicate the incentive effects, δ 1j (j = 1,2,3) measures the pre-expiration

effect, δ 2 measures the post-expiration effect, and δ 3 indicates the size of the spike in the

month benefit expire. Note that we can identify δ 1 because of the “natural experiment”

character of the change in the benefit law. Note too that we can identify δ 2 because

benefits expire at different durations of unemployment. If they all expired at the same

duration we could not distinguish the post-expiration effect from duration dependence.

The same holds for the spike. We can identify the spike because it occurs at different

unemployment durations. Otherwise, we could not distinguish the spike from the effect of

duration dependence. 8

The conditional density function of the completed unemployment duration t e that ended

in a transition toward a job can be written as

f(t e | x) = θ(t e | x) exp(−

∫ te

0

θ(s | x)ds) (4)

Since we analyze an inflow sample, the log-likelihood L of the model is rather straightforward,

consisting of two components

L = dΣlog(f) + (1 − d)Σlog(1 − F ) (5)

8 Also note that because of the δ parameters duration dependence is different for different individuals,

so the specification is a nonproportional hazard type.

17


where F is the distribution function of f and d is a dummy variable with a value of 1 if the

worker found a job and a value of 0 if the worker is still unemployed or left unemployment

for other reasons. The transition rate to other destinations θ n (t | x) can be modeled

similarly. Then, those who have found a job are assumed to have a right-censored duration

with regard to exits to other destinations. Note that we can estimate the parameters of

the job-finding rate and the exit rate to other destinations separately because they are

assumed to be independent. 9

5.3 Parameter estimates

The parameters of the model are estimated separately for males and females using the

method of Maximum Likelihood (see table 4). Many of the parameters representing incentive

effects are significantly different from zero. Only the coefficient attached to the

indicator of a reduction of PBD from 6 to 3 months does not differ significantly from

zero in the job-finding rate. The similar parameter in the exit rate to other transitions

does differ significantly from zero. Apparently, reducing the benefit period to 3 months

provides the unemployed with insufficient time to find a job more quickly. The outflow

from unemployment is stimulated, but only to other destinations, such as active labor

market policy programs. The other PBD reductions have significant effects both on the

job-finding rates and on the transition rates to other destinations. There are also clear

post-benefit expiration effects. When unemployed workers lose their benefits, the outflow

from unemployment increases, both through higher job-finding rates and through higher

exit rates to other destinations. Finally, the parameter estimates show clear spikes in the

month unemployment benefits expire.

Neither for males nor for females do changing conditions in the labor market seem to

affect the job-finding rate but they do affect exits to other destinations. Perhaps this has to

do with changes in active labor market policies that made it easier for unemployed workers

to enter labor market programs.

9 We investigate the validity of this assumption below.

18


Other results are also interesting. Age has a negative effect on the exit rate from unemployment,

both through job finding and other exits. Education does not have a significant

effect for males, but females in the highest educational category have a significantly higher

exit rate than those in other educational categories. Males in the higher education categories

have a lower exit rate to other destinations, but the opposite is true for females. The

effect of family conditions also differs for males and females. Males who have dependent

family members have a higher job-finding rate than males who do not, but females with

dependent family members have a lower job-finding rate than other females. Apparently

having dependent family members is a stimulus for males to leave unemployment more

quickly but a handicap for females. Bad health reduces the exit rate from unemployment

substantially. Work experience has a positive effect on the job-finding rate, but does not

affect exits to other destinations.

The pattern of duration dependence for job-finding rates is different from that for exit

rates to other destinations. There is negative duration dependence for the job-finding

rates 10 and positive duration dependence for the exit rates to other destinations. Apparently,

as the period of unemployment progresses it becomes increasingly difficult to find a

job, but easier to leave unemployment for other reasons.

5.4 Sensitivity analysis

We performed several sensitivity analyses to investigate the sensitivity of the parameter estimates

for model specifications. First, we expanded the model by introducing unobserved

heterogeneity terms in the job-finding rate and the exit rate to other destinations, allowing

for correlation between these terms. And the estimation results did indeed indicate the

presence of correlated unobserved heterogeneity. 11 But the introduction of unobserved heterogeneity

hardly affects the other parameter estimates for either males of females, so our

10 For males there is a peak in the job-finding rate in the third month. This peak could have to do with

the replacement rate dropping from 70% to 60% after three months. For females we do not find a similar

effect.

11 The parameter estimates are available on request.

19


main conclusions about the incentive effects of the reduction in PBD remain unchanged.

We also investigated the specification of duration dependence. Instead of using quarterly

and half-yearly duration intervals for longer unemployment durations we used monthly

duration intervals for up to 2 years. This gave more detailed information about the pattern

of duration dependence but did not affect the estimates of the relevant parameters – that

is, the incentive effects.

Moreover, we examined whether age has a nonlinear effect on the job-finding rate

and the exit rate to other destinations.

We replaced age as a continuous variable by

dummy variables representing age categories of 5 years. This did not change the parameter

estimates much and had no effect on the relevant benefit duration variables.

Finally, we investigated whether the size of the end-of-benefit spikes depended on calendar

time (before and after the law changed) or group of workers (control group or treatment

group). It turns out that the end-of-benefit spikes after the law changed are somewhat

smaller than those before it changed. 12 However, estimates of the pre-expiration and postexpiration

effects do not change if we allow the after-end-of-benefit spikes to be different

from the before-end-of-benefit spikes. There are also differences among groups in the endof-benefit

spikes – that is, the spike increases with work experience. But estimates of the

pre-expiration and post-expiration effects are not affected by a more flexible specification

of the end-of-benefit spikes.

To indicate the size of the effects of reform, we calculated the difference in exit rates

before and after the law changed for a median unemployed worker, a 30-year-old male in

good health, with vocational school education, no dependents, and work experience of 10

to 15 years. 13

Such a worker had a 12-month benefit entitlement before the law changed

and a 6-month benefit entitlement after. The effects are dramatic (table 5). Before the

law changed, that worker had a 49 percent probability of finding a job within 6 months

12 Only for the job-finding rates of males do we not find significant differences between the before- and

after-end-of-benefit spikes.

13 The average age of an unemployed male worker in our sample was 31.2 years. Of the males, 41.2%

had vocational school education, 60% had no dependent family members, 92.3% were in good health, and

24.6% had work experience of 10 to 15 years.

20


of becoming unemployed, and a 4 percent probability of leaving unemployment for other

reasons. The corresponding percentages after 12 months were 65.8 (exit to employment)

and 9.3 (exit for other reasons). After the unemployment benefit law changed, the worker’s

probability of finding a job within 6 months jumped to 58.8 percent (an increase of 9.5

percentage points) and the probability of leaving unemployment for other reasons rose to

10.7 percent (an increase of 6.4 percentage points). Thus, the overall probability of workers

in this group leaving unemployment within 6 months increased from 53.6 percent in the

period before the law changed to 69.5 percent after. Another way of viewing the faster exit

from unemployment is to compare job-finding rates 12 months into unemployment spell.

After 12 months 72.5 percent had found a job and 15 percent had left unemployment for

other reasons. The median unemployment duration for the median worker fell from 5.3

months before the law changed to 4.1 months after. On average his means that for every

week cut from potential benefits unemployment was cut by about 1.3 days, which is more

in other studies we cited.

For a 40-year-old in the same reference group, the job-finding probabilities are less

but are still higher than before the change in the unemployment law (see table 5). Both

the job-finding rates and the exit rates to other destinations are substantially smaller for

women in the same reference group, only 57.3 percent have left unemployment within 6

months, compared with 69.5 percent for their male counterparts. The median duration of

unemployment for women in this group fell from 8.9 months before the law changed to 5.4

months after.

6 Conclusions

In our analysis on job-finding and other exits destinations from unemployment, we found

important and sizeable disincentive effects of the unemployment insurance system. First,

we identified clear spikes in job finding at the points when benefit were exhausted. Second

and most important, the probability of finding a job greatly increased for most groups of

recipients whose benefit entitlement period was shortened, but stayed virtually unchanged

21


for recipients whose entitlement period did not change. Third, the rate of job finding after

benefit expiration was substantially higher than before (excluding the spike in the month of

benefit expiration). We found similar effects on the exit rate from unemployment to other

destinations, including active labor market programs. Apparently policymakers were able

to attract benefit recipients to active labor market programs, but more intense employment

services and monitoring may also have contributed to this development.

Did workers find jobs faster because they faced a shorter period of benefit entitlement

or because more intense employment services were offered to the unemployed and was

monitoring stricter in the period after the law changed? The fact that job-finding rates

did not change for the unemployed workers whose entitlement period did not change, and

that spikes shifted as they did in the hazard rates for the treatment groups, suggest that

the job-finding rate improved mainly because of more intense job-search efforts by the

unemployed workers. Moreover, younger workers are particularly sensitive to changes in

the business cycle, so their relative performance after reform should have been better than

for other groups, but it was worse, which is again further evidence of a causal effect.

What lessons can be learned from the change in Slovenia’s unemployment benefit law?

The reform was certainly effective at encouraging benefit recipients to leave unemployment,

thus reducing the severity of the moral hazard induced by the unemployment benefit system.

Those positive developments must be weighed against additional hardships created

by the curtailment of benefits and against the possible lower quality of post-unemployment

jobs in terms of stability, type of appointment, and precariousness. For a thorough assessment

of the legislative changes, it is important to probe qualitative issues as well – an

important area for future research.

22


References

[1] Atkinson, A.B., and J. Micklewright (1991) Unemployment Compensation and Labor Market

Transitions: A Critical Review, Journal of Economic Literature 29 (4): 1679-1727.

[2] Card, D.E. and P.B. Levine (2000) Extended Benefits and the Duration of UI Spells: Evidence

from the New Jersey Extended Benefit Program, Journal of Public Economics 78: 107-138.

[3] Carling, K., P-A. Edin, A. Harkman and B. Holmlund (1996) Unemployment Duration, Unemployment

Benefits, and Labor Market Programs in Sweden, Journal of Public Economics

59: 313-334.

[4] Katz, L.F., and B.D. Meyer (1990) The Impact of the Potential Duration of Unemployment

Benefits on the Duration of Unemployment, Journal of Public Economics 41 (1): 45-72.

[5] Lalive, R. and J. Zweimüller (2004) Benefit Entitlement and Unemployment Duration: Accounting

for Policy Endogeneity, Journal of Public Economics 88(12): 2587-2616.

[6] Lalive, R., J.C. van Ours, and J. Zweimüller (2004) How Changes in Financial Incentives

Affect the Duration of Unemployment, CentER Discussion Paper, 04-86, Tilburg University.

[7] Mortensen, D.T. (1977) Unemployment Insurance and Job Search Decisions, Industrial and

Labor Relations Review 30 (4): 505-517.

[8] Roed, K. and T. Zhang (2003) Does Unemployment Compensation Affect Unemployment

Duration? Economic Journal 113: 190-206.

[9] Van den Berg, G.J. (2001) Duration Models: Specification, Identification, and Multiple Durations,

in: Heckman, J.J. and Leamer, E. (eds), Handbook of Econometrics, Volume V,

North-Holland.

[10] Van Ours, J.C. and M. Vodopivec (2004) How Changes in Benefits Entitlement Affect jobfinding:

Lessons from the Slovenian “Experiment,” IZA Discussion Paper No. 1181.

23


Table 1 Cumulative probability of exits to a job or other destination, before and

after the 1998 change in Slovenia’s unemployment benefits law (as percentage, by

duration of unemployment) a)

Before change of law After change of law Increase of outflow

Duration of Job Other Total Job Other Total Job Other Total

unemployment (%) (%) (%) (%) (%) (%) (%) (%) (%)

Males

≤ 3 months 28.7 1.2 29.9 31.2 5.3 36.5 2.5 4.1 6.6

≤ 6 months 45.8 3.3 49.1 51.0 12.0 63.0 5.2 8.7 13.9

≤ 9 months 54.9 5.5 60.4 59.4 15.7 75.1 4.5 10.2 14.7

≤ 12 months 61.6 8.2 69.8 63.5 18.1 81.6 1.9 9.9 11.8

Females

≤ 3 months 21.8 1.3 23.1 25.5 6.5 32.0 3.7 5.2 8.9

≤ 6 months 35.8 3.2 39.0 42.0 13.9 55.9 6.2 10.7 16.9

≤ 9 months 45.7 5.6 51.3 50.7 18.4 69.1 5.0 12.8 17.8

≤ 12 months 53.5 8.4 61.9 55.7 21.0 76.7 2.2 12.6 14.8

a) The calculations are based on a sample of 9,196 males and 10,853 females.

24


Table 2 How changes in the duration of benefit entitlements affected the duration of

unemployment

Maximum duration Median duration of

Experience of benefits (months) unemployment (months)

Group (years) Before After Age a) Before After Diff.

1 0-1.5 3 3 19-29 4.9 4.2 -0.7

2 1.5-5 6 3 21-30 5.6 4.2 -1.4

3 5-10 9 6 23-35 7.7 5.1 -2.6

4 10-15 12 6 27-39 9.4 5.3 -4.1

5 15-20 18 9 32-43 12.1 6.4 -5.7

a) The age boundaries are determined by the presence of at least 100 observations for a particular

year of age.

25


Table 3 Probability of leaving unemployment within a particular period of time;

parameter estimates logit models a)

Probability of leaving unemployment

Males ≤ 3 months ≤ 6 months ≤ 9 months ≤ 12 months

After change of law 0.14 (1.4) 0.38 (3.5)* 0.33 (2.6)* 0.34 (2.4)*

Age/10 -0.33 (5.0)* -0.44 (7.2)* -0.59 (9.2)* -0.66 (9.6)*

Education2 -0.21 (2.0)* -0.06 (0.6) 0.03 (0.3) -0.03 (0.3)

Education3 -0.31 (3.2)* -0.18 (1.8) -0.02 (0.2) -0.02 (0.2)

Education4 -0.51 (4.9)* -0.43 (4.2)* -0.30 (2.7)* -0.29 (2.4)*

Family1 0.03 (0.5) 0.13 (2.0)* 0.13 (1.9) 0.15 (2.0)*

Family2 0.02 (0.3) 0.12 (2.0)* 0.10 (1.6) 0.12 (1.8)

Ill health -1.36 (12.2)* -1.57 (17.6)* -1.70 (20.0)* -1.81 (21.8)*

Experience 1.5-5 years -0.56 (4.2)* -0.23 (1.8) -0.12 (0.9) -0.22 (1.4)

Experience 5-10 years -0.24 (2.3)* -0.33 (3.2)* -0.24 (2.1)* -0.16 (1.2)

Experience 10-15 years -0.28 (2.4)* -0.33 (2.9)* -0.44 (3.7)* -0.10 (0.8)

Experience 15-20 years -0.05 (0.3) -0.22 (1.6) -0.32 (2.3)* -0.33 (2.2)*

Incentive effects

PBD 6 to 3 months 0.58 (3.5)* 0.31 (1.9) 0.28 (1.5) 0.39 (1.8)

PBD 9 to 6 months 0.10 (0.7) 0.28 (1.9) 0.30 (1.9) 0.19 (1.0)

PBD 12 to 6 months 0.17 (1.2) 0.34 (2.4)* 0.59 (3.8)* 0.32 (1.8)

PBD 18 to 9 months 0.09 (0.7) 0.08 (0.6) 0.50 (3.2)* 0.64 (3.7)*

Constant 0.77 (3.9)* 1.86 (9.8)* 2.75 (13.5)* 3.36 (15.0)*

26


Probability of leaving unemployment

Females ≤ 3 months ≤ 6 months ≤ 9 months ≤ 12 months

After change of law 0.26 (2.8)* 0.43 (4.8)* 0.40 (4.2)* 0.48 (4.4)*

Age/10 -0.18 (3.0)* -0.30 (5.3)* -0.36 (6.4)* -0.44 (7.4)*

Education2 -0.06 (0.4) -0.07 (0.6) -0.08 (0.6) -0.08 (0.6)

Education3 -0.23 (1.7) -0.20 (1.5) -0.23 (1.8) -0.11 (0.8)

Education4 -0.06 (0.4) -0.01 (0.0) 0.06 (0.5) 0.22 (1.7)

Family1 -0.15 (2.7)* -0.19 (3.5)* -0.21 (3.8)* -0.19 (3.2)*

Family2 -0.14 (2.4)* -0.19 (3.6)* -0.26 (4.8)* -0.18 (3.3)*

Ill health -2.04 (10.6)* -1.86 (14.6)* -1.89 (17.2)* -1.89 (19.0)*

Experience 1.5-5 years -0.33 (2.8)* -0.10 (1.0) 0.03 (0.3) 0.09 (0.7)

Experience 5-10 years -0.26 (2.5)* -0.34 (3.7)* -0.04 (0.5) 0.02 (0.2)

Experience 10-15 years -0.26 (2.3)* -0.41 (4.1)* -0.33 (3.3)* -0.09 (0.8)

Experience 15-20 years -0.17 (1.2) -0.28 (2.4)* -0.25 (2.1)* -0.30 (2.4)*

Incentive effects

PBD 6 to 3 months 0.36 (2.5)* 0.10 (0.7) 0.13 (0.9) -0.02 (0.1)

PBD 9 to 6 months 0.07 (0.5) 0.25 (2.0)* 0.25 (1.9) 0.10 (0.7)

PBD 12 to 6 months 0.25 (1.9) 0.51 (4.2)* 0.56 (4.4)* 0.23 (1.6)

PBD 18 to 9 months 0.19 (1.4) 0.24 (1.9) 0.60 (4.6)* 0.61 (4.3)*

Constant -0.20 (0.9) 0.97 (4.8)* 1.60 (7.8)* 2.09 (9.7)*

a) Samples of 9,196 males and 10,853 females;

absolute t-statistics in parentheses;

a * indicates a 95% significance level.

27


Table 4 Parameter estimates for hazard rate models

Males

Females

To job To other To job To other

Incentive effects

PBD reduced to 3 months 0.00 (0.0) 0.40 (2.1)* -0.03 (0.4) 0.46 (3.0)*

PBD reduced to 6 months 0.18 (3.4)* 0.82 (7.4)* 0.23 (4.3)* 0.93 (9.8)*

PBD reduced to 9 months 0.27 (4.1)* 0.89 (6.9)* 0.36 (5.7)* 0.99 (8.9)*

After benefits expiration 0.53 (10.2)* 0.70 (7.1)* 0.43 (8.8)* 0.92 (10.2)*

End of benefits spike 0.82 (16.7)* 1.14 (11.2)* 0.91 (18.9)* 1.16 (13.0)*

After change of law -0.07 (1.7) 0.20 (3.2)* -0.05 (1.6) 0.21 (4.2)*

Characteristics

Age/10 -0.37 (9.9)* -0.24 (3.6)* -0.33 (9.7)* -0.13 (2.4)*

Education2 0.01 (0.1) -0.22 (2.2)* 0.04 (0.5) 0.37 (2.9)*

Education3 0.05 (0.9) -0.44 (4.6)* 0.04 (0.5) 0.05 (0.4)

Education4 -0.09 (1.4) -0.51 (5.1)* 0.19 (2.3)* 0.36 (2.9)*

Family1 0.09 (2.5)* -0.02 (0.3) -0.06 (2.0)* -0.08 (1.6)

Family2 0.10 (3.0)* -0.11 (1.7) -0.10 (3.4)* -0.09 (1.9)

Ill health -1.33 (22.0)* -1.29 (14.4)* -1.52 (20.1)* -1.25 (13.5)*

Experience 1.5-5 years 0.07 (1.5) -0.02 (0.3) 0.08 (1.8) 0.12 (1.7)

Experience 5-10 years 0.20 (3.9)* -0.07 (0.7) 0.13 (2.9)* 0.04 (0.6)

Experience 10-15 years 0.30 (5.0)* 0.07 (0.6) 0.22 (4.0)* 0.04 (0.4)

Experience 15-20 years 0.38 (4.9)* -0.08 (0.6) 0.35 (4.9)* -0.08 (0.7)

28


Males

Females

To job To other To job To other

Duration dependence

Duration Month 2 0.08 (1.7) 0.20 (1.2) -0.03 (0.7) 0.32 (2.5)*

Duration Month 3 0.20 (4.0)* 0.39 (2.5)* -0.20 (3.6)* 0.30 (2.3)*

Duration Month 4 0.06 (1.0) 0.99 (6.8)* -0.24 (4.3)* 0.63 (5.0)*

Duration Month 5 0.00 (0.1) 0.88 (5.8)* -0.18 (3.3)* 0.61 (4.7)*

Duration Month 6 -0.23 (3.7)* 0.67 (4.0)* -0.40 (6.4)* 0.33 (2.3)*

Duration Months 7-9 -0.46 (8.1)* 0.97 (6.8)* -0.49 (8.9)* 0.63 (5.0)*

Duration Months 10-12 -0.63 (9.3)* 1.21 (7.9)* -0.55 (8.8)* 0.69 (5.0)*

Duration Months 13-18 -0.79 (11.3)* 1.21 (7.9)* -0.84 (12.6)* 0.73 (5.3)*

Duration Months 18+ -1.74 (20.4)* 0.96 (6.0)* -1.51 (19.8)* 0.62 (4.5)*

Constant -4.83 (43.2)* -7.43 (33.3)* -5.09 (40.0)* -8.37 (37.9)*

–Loglikelihood 45,453.4 15,567.4 51,682.5 22,303.5

a) Samples of 9,196 males and 10,853 females;

absolute t-statistics in parentheses;

a * indicates a 95% significance level.

29


Table 5 Changes in the cumulative probability of leaving unemployment (as a %)

and in median duration of unemployment (in months); simulation results

Before After Median duration

PBD 12 months PBD 6 months Difference (months)

After Job Other Total Job Other Total Job Other Total Before After Diff.

Median worker a) 5.3 4.1 -1.1

6 months 49.3 4.3 53.6 58.8 10.7 69.5 9.5 6.4 15.9

12 months 65.8 9.3 75.1 72.5 15.0 87.5 6.7 5.7 12.4

If age = 40 8.5 5.4 -3.1

6 months 37.7 3.8 41.5 46.7 10.1 56.8 9.0 6.3 15.3

12 months 53.3 9.3 62.5 61.7 15.5 77.2 8.4 6.2 14.6

If female 8.9 5.4 -3.5

6 months 36.1 3.5 39.6 47.3 10.0 57.3 11.2 6.5 17.7

12 months 55.2 7.4 62.6 62.8 14.8 77.6 7.6 7.4 15.0

a) 30-year-old male in good-health, with vocational school education, 10 to 15 years of work

experience, and no dependent family members. Before the change in the benefit law this person

was entitled to 12 months of unemployment benefits; after the change, he was entitled to 6

months. Simulation results are based on the parameter estimates presented in table 4.

30


Figure 1: Survival in unemployment, before and after the law changed (by entitlement

group)

a. Eligibility 3 months before - 3 months after

b. Eligibility 6 months before - 3 months after

Survival (%)

100

90

80

70

60

50

40

30

20

10

0

0 5 10 15 20 25

Survival (%)

100

90

80

70

60

50

40

30

20

10

0

0 5 10 15 20 25

Months of unemployment

Months of unemployment

Before

After

Before

After

c. Eligibility 9 months before - 6 months after

d. Eligibility 12 months before - 6 months after

Survival (%)

100

90

80

70

60

50

40

30

20

10

0

0 5 10 15 20 25

Survival (%)

100

90

80

70

60

50

40

30

20

10

0

0 5 10 15 20 25

Months of unemployment

Months of unemployment

Before

After

Before

After

e. Eligibility 18 months before - 9 months after

Survival (%)

100

90

80

70

60

50

40

30

20

10

0

0 5 10 15 20 25

Months of unemployment

Before

After

1

31


Figure 2: Monthly job-finding rates, before and after the law changed (by entitlement

group)

a. Eligibility 3 months before - 3 months after

b. Eligibility 6 months before - 3 months after

0.25

0.25

0.2

0.2

Monthly exit rate

0.15

0.1

Monthly exit rate

0.15

0.1

0.05

0.05

0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Months of unemployment

Months of unemployment

Before

After

Before

After

c. Eligibility 9 months before - 6 months after

d. Eligibility 12 months before - 6 months after

0.25

0.25

0.2

0.2

Monthly exit rate

0.15

0.1

Monthly exit rate

0.15

0.1

0.05

0.05

0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Months of unemployment

Months of unemployment

Before

After

Before

After

e. Eligibility 18 months before - 9 months after

0.25

0.2

Monthly exit rate

0.15

0.1

0.05

0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Months of unemployment

Before

After

1

32


Figure 3: Monthly exit rates to other destinations, before and after the law changes (by

entitlement group)

a. Eligibility 3 months before - 3 months after

b. Eligibility 6 months before - 3 months after

0.25

0.25

0.2

0.2

Monthly exit rate

0.15

0.1

Monthly exit rate

0.15

0.1

0.05

0.05

0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Months of unemployment

Months of unemployment

Before

After

Before

After

c. Eligibility 9 months before - 6 months after

d. Eligibility 12 months before - 6 months after

0.25

0.25

0.2

0.2

Monthly exit rate

0.15

0.1

Monthly exit rate

0.15

0.1

0.05

0.05

0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Months of unemployment

Months of unemployment

Before

After

Before

After

e. Eligibility 18 months before - 9 months after

0.25

0.2

Monthly exit rate

0.15

0.1

0.05

0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Months of unemployment

Before

After

1

33


Figure 4:

Changes in potential duration of benefits and in the job-finding rate; preexpiration

and post-expiration effects

θ

θ A

δ 1

δ 2

34

T A

θ B

T B

t

More magazines by this user
Similar magazines