The probability of reemployment and the duration of unemployment

percentage points³⁴ more likely to be working a quarter later. For both sexes, the probability of finding employment increases significantly as qualifications become higher and decreases with the time since last in employment. As expected, the presence of small children in the family affects the behaviour of men and women differently: the presence of a child below years of age made men 3.6 percentage points more likely and women 8.5 percentage points less likely to take up employment. The estimated effects are not very substantial, but taking into account the average probability of employment (18.7%), they are not insignificant either. For instance, the fact that no one works in the family reduces the average probability of finding a job by 5.4 percentage points, i.e., 0.054/0.186=27%.

5.4. The probability of reemployment and the duration of

Table 17: Persons exhausting their eligibility for UI in the 3rd quarter and the number of quarters they spent in the observation

Number of quarters

in the sample Male Female Total

2 190 153 343

3 168 105 273

4 135 69 204

5 55 47 102

Total 548 374 922

As compared to the total group of unemployed, the probability of finding employment in the next quarter is much smaller among this group: instead of the pervious 18.98 and 18.34%, now it is 7.88 and 6.46% (Table 18).

Table 18: Ratio of unemployed exhausting eligibility for UI who find work, by sex

Males Females Total

No 92,12% 93,54% 92,69%

Will he/she be employed

in the next quarter? Yes 7,88% 6,46% 7,31%

total 100% 100% 100%

As, like in the previous model, we also examined the probability of taking up employment, we considered more or less the same factors, except for the variables indicating the durations.

However, of these factors, we left in the model only those the effects of which turned out to be significant or proved significant together with other variables (based on an F-test). We estimated the effects of the variables using several³⁵ duration models, which yielded very similar results.

35 multi-stage and continuous time, parametric and non-parametric models. Of these, we disclose the Jenkins estimates, which is an estimated using a multi-stage model and logit function.

Table 19: Average effect of various factors on the hazard of taking up employment

Males Females

Becoming employed APE p-value APE p-value

RSA -0,0596 0,005 -0,0557 0,077

Public work - - - -

Quarter 1 0,0014 0,953 -0,0506 0,080

Quarter 2 0,0072 0,730 0,0153 0,539

Quarter 3 -0,0299 0,095 -0,0174 0,459

18-24 years old 0,0990 0,088 0,0827 0,000

25-34 years old 0,0676 0,171 0,0647 0,000

35-54 years old 0,0790 0,085 0,0406 0,000

Elementary school 0,0152 0,666 0,0123 0,784 Technical school/skilled

worker 0,0321 0,433 0,0077 0,878

Secondary school - - -0,0159 0,709

Vocational secondary

school 0,0422 0,368 0,0425 0,494

Working spouse 0,0236 0,209 -0,0027 0,891 Household with child(ren) 0,0109 0,560 -0,0011 0,967 Unemployment rate of the

region -0,2678 0,017 -0,2318 0,159

Number of months since

registrated unemployed -0,0111 0,000 -0,0104 0,002 Number of months since

registrated unemployed (squared term)

0,0001 0,000 0,0001 0,000

t2 -0,1976 0,010 -0,5727 0,000

t3 -0,1597 0,003 -0,3593 0,000

t4 -0,1266 0,000 - -

t5 -0,1190 0,002 -0,3512 0,000

Year 1 0,0981 0,064 0,1438 0,066

Year 2 0,0633 0,068 0,0957 0,066

Year 3 -0,0048 0,924 0,0564 0,461

Number of observations 1 023 607

Prob> chi2: 0,000 0,000

Notes: The estimates were made using a logit function, where the dependent variable was the probability of taking up employment. Variables significant at the 10% level are indicated in bold letters. The control group was the 55-62-year-old age group for age, and persons with no more than elementary education for school qualification.

Source: Own calculations based on the 2001-2004 Labour Force Survey.

In this group, the effect of public work could not be estimated statistically because, of the small number of unemployed in public work (51 out of 1053 observed men, 31 out of 607 women), no one

took up employment. In contrast, the effect of the regular social assistance is significant and negative: benefit recipient men are 6.9 percentage points, women 5.6 percentage points less likely to take up employment than their non-recipient counterparts, taking into consideration also the time since the exhaustion of eligibility to the UI. The size of the effect must be viewed in light of the fact that the likelihood of taking up employment in the quarter after the observation is on average 7.9% for men and 6.5% for women. That is, men recipients of regular social assistance have one quarter³⁶, women 85% less chance to become employed than their non-recipient peers. Though to a lesser extent, but it is still a valid concern that recipients may represent a special group within the examined persons, which, as a result, may also reflect the effects of non-observed factors other than the disincentive effect. That is because, even though in certain respects of their labour market links and work history, we have made the observed group homogeneous, this has not made the recipients any more similar to the non-recipients in terms of the probability of taking up employment.

As compared to the model described in Table 16, there are fewer variables with significant impacts in the duration model. One reason may be the much smaller number of elements, but it may also be attributable to the greater homogeneity of the group. In contrast to the qualifications and the characteristics of the household, age does have an effect on the probability of employment: younger males were 7-10 percentage points more likely to find employment, and women 4-8 percentage points more likely, than the age cohort above 55. The past duration of unemployment and, for males, the unemployment rate of the region had a negative effect on the probability of employment.

Using the probabilistic hazards, the duration of staying unemployed after entering the scope of observation can be calculated; adding that to the time of unemployment before the observation period, we can also estimate the total duration of unemployment after the loss of the latest job (Table 20). Based on those estimates we can conclude that among the unemployed persons who have just exhausted their eligibility to UI, those who receive RSA remain unemployed 7 quarters (almost 2 years) longer than non-recipients.

36 Thus, benefit recipient men have a chance of 7.9%-5.9%=2%, and women, 6.5%-5.5%-1%, to be employed a quarter later.

Table 20: Average duration of unemployment in quarters

Mean St. error Number of observations

no 9,77 5,96 567

Does he/she

receive RSA? yes 16,16 7,24 355

Total 12,23 7,19 922

Source: Own calculations based on the 2001-2004 Labour Force Survey.