HRDC - The Impact of Employment Insurance on New-entrants and Re-entrant Workers

Directorates

· Director General's Office

· Data Development

· Strategic Evaluations

· Partnership and Continuous Evaluations

Publications

· Evaluation Reports

· Evaluation Briefs

· Labour Market Development Agreement Series

· Lessons Learned Series

· Evaluation Tool Kit Series

· Evaluation Forum Newsletter

· From Theory To Practice

4. How Has EI Affected NERE Access to Benefits?

To address the question, “does the switch from Unemployment Insurance (UI) to Employment Insurance (EI) reduce new and re-entrants (NEREs)’ access to unemployment insurance benefits?”, we start with an analysis of the incidence of receiving benefits for all NEREs and then for various sub-populations of NERE workers. Secondly, we analyze the probability of NEREs receiving benefits using multivariate techniques.

4.1 Descriptive Analysis

Note first that NEREs are much less likely to receive benefits than non-NEREs (recipiency rates are generally one half) under either UI or EI. Overall, there was a very small decline in the percentage of NEREs reporting benefits as a result of the full switch from UI to EI. For example, 26.9 percent of NEREs with job separation prior to July 1996 reported receiving unemployment insurance benefits and 23.8 percent of NEREs with job separation after January 1997 reported receiving unemployment insurance benefits (see Table 4).⁹ The change in the percentage of NEREs reporting benefits potentially can reflect (i) the increase in eligibility requirements from 20 to 26 weeks and (ii) the change from a weeks-based system to an hours-based system. To distinguish the relative impacts of these two changes we present results for the 3 time periods: (i) before July 1996, the date at which the eligibility condition was increased from 20 to 26 weeks; (ii) the interim period from July 1996 to December 1996 during which the increased eligibility condition was in place, but benefits still depended on the number of weeks of work during the qualifying period; and (iii) the period after January 1997 when the switch from weeks to hours was made.¹⁰

By distinguishing the effects of the two policy changes we can see that the increase in eligibility conditions substantially reduced the access to benefits for NERE workers (26.9 percent of NERE separators received benefits prior to July 1996 versus only 17.8 percent after July 1996), but the switch from a weeks-based system to an hours-based system significantly improved this situation (from 17.8 percent recipiency to 23.8 percent recipiency). Thus, the net effect of the two changes is only a small reduction in the overall rates of recipiency for NEREs.

Notice that there is a similar overall decline in the access to benefits among regular workers. For example, non-NEREs with benefits as a percentage of all job separators with benefits, declined from 48.9 to 45.5 percent (see Table 4, row 1). However, there is not the same dip in benefit receipt between the pre-July 1996 and July-December 1996 periods for non-NERE separators (nor did they experience the same significant increase in eligibility conditions).

This basic pattern of change in the access to benefits observed for all NEREs can also be found for various socio-economic groups considered separately, with some exceptions and these are reported here. Results on NEREs reporting benefits as a percentage of all NEREs, for various categories, are presented in Table 5B and for comparison, results for all non-NEREs are presented in Table 5A. Benefit receipt among NEREs declined in each of the age groups considered, with the exception of the youngest age group (15-24 years) in which case, the percentage of these youngest age NEREs receiving benefits remained about the same (at very low levels; only about 15 percent of young NEREs received benefits). The largest reduction in benefit receipt occurred for the 35 to 44 age groups (from 39.2 percent to 32.4 percent). For all but the oldest age group, benefit recipiency declined as a result of the tighter eligibility condition, then recovered somewhat with the switch from a weeks-based to an hours-based system.

The results by gender show that while the percentage of NEREs receiving benefits decreased for women (from 27.6 to 20.6 percent), it actually increased very slightly for men (from 26.3 to 27.1 percent). Notice that female NEREs experienced a larger drop in eligibility than men as a result of the increase in minimum eligibility conditions from 20 to 26 weeks. If we look at NEREs with children living at home, it is clear that benefit receipt dropped quite strikingly for NERE married mothers, fell for NERE single mothers, and increased for NERE fathers. We also looked at mothers and fathers with a youngest child aged 6 to 10 (our potentially re-entering parent case). Again, we found that recipiency rates fell dramatically for mothers, but increased for fathers.

NEREs in some provinces experienced increases in access to benefits, namely, Newfoundland, New Brunswick, and Quebec. In terms of changes in benefit receipts for different levels of income, there was an increase in benefit receipt among NEREs in households with income levels in the $35,000-49,999 range, whereas, for all-non NERE households, benefit receipts fell in each of the household income categories.

4.2 Multivariate Analysis

We also examined whether the policy changes resulted in a decline in access to (un)employment insurance benefits for NERE workers, after controlling for worker characteristics and some employment characteristics. We first estimated a probit model of the probability of receiving benefits for all workers (NEREs and non-NEREs) with job separation. The key explanatory variables are: (i) a dummy variable (WEEKS26) which indicates whether the individual’s job separation occurred between July 1996 and December 1997 and captures the impact of the increase in the number of weeks from 20 to 26 of required employment to qualify for benefits; (ii) a dummy variable (HRSBASE) which indicates whether the individual’s job separation occurred after January 1997 and captures the impact of the switch from a weeks-based to an hours-based criterion¹¹; and (iii) a dummy variable (NERE) which indicates whether the individual is a NERE worker. (i) and (ii) are also entered interactively with (iii) to assess whether NEREs disproportionately experience a decrease in access to unemployment insurance benefits in the two EI periods, compared to the UI period. We also include a dummy variable (NEW) which indicates that the worker is a new entrant,¹² a set of seasonal, age, education, marital status, and children dummy variables, the hourly wage at the time of the separation, household equivalent income¹³, and the regional unemployment rate.¹⁴

A second version of this model is estimated with gender and youth dummy variables interacted with both NEREs and the policy change variables since we have argued that these two groups may be particularly affected. Results are presented in Table 6.

A first main result from the probit analysis using all job separators is that the NERE variable is negative and significant indicating that the probability of a NERE worker receiving benefits is significantly lower than the probability of a regular worker receiving benefits [see Table 6, column 1]. Notice also that the NEW variable is negative and significant indicating that the probability of receiving benefits for a new entrant is even lower than that of the other NERE workers.¹⁵

In terms of policy effects, we find that the WEEKS26 variable is insignificant overall, which is reasonable since non-NERE workers are unaffected by this change aimed specifically at NEREs. However, the HRSBASE variable is negative and significant, indicating that for all workers, the switch from the weeks to the hours-based criterion reduced benefit eligibility. The interaction of the NERE variable with the WEEKS26 variable is negative and significant indicating that the tighter eligibility criterion reduces the probability of a NERE worker receiving benefits. However, the interaction of the NERE and the HRSBASE variable is positive and significant indicating that the switch from the weeks to hours-based criterion increases the probability of a NERE worker receiving benefits. Thus, the 2 policy changes had off-setting impacts on NEREs overall, confirming the story apparent in the descriptive tables discussed earlier.

The second version of this model (Table 6, column 2) shows that female NERE workers are less likely than male NERE workers to receive benefits (though the opposite is true for non-NEREs). Notice that while both the female and male NERE variables (NEREFEM and NEREMAL) are negative, the coefficient on the female NERE worker variable is larger. Both the male and female NERE workers have lower probabilities of receiving benefits after the increase in the number of weeks-worked criterion (compared to the original criterion) since the interactions of NEREFEM and WEEKS26 and NEREMAL and WEEKS26 variables are negative and significant, though the reduction is larger for female NEREs. Finally, for both the male and female NERE workers, the switch to the hours-based criterion was beneficial, bringing the probability of receiving benefits back up almost to the pre-July 1996 level. The magnitude of this softening effect is roughly the same for men and women. Thus, it was the larger effect on women of the increase to a 26-week entrance requirement which results in female NEREs being worse off overall.

Younger NERE workers (15 to 24 years) are also more affected by the policy changes than others, especially by the increase in the minimum required weeks. In fact, the policy impact on NERE youth is the largest observed.

Having examined the probability of all workers (regular and NERE workers combined) receiving (un)employment insurance benefits, we turn next to focus upon a more detailed analysis of the determinants of the probability of NERE workers alone receiving benefits (see Table 7). We estimate a probit model of NERE workers receiving (un)employment insurance benefits where the two key explanatory variables included to capture the policy changes are again, (i) a dummy variable WEEKS26 indicating whether the individual’s job separation occurred when the criterion increased from 20 to 26 weeks of employment (the period July to December 1996) and (ii) a dummy variable HRSBASE indicating whether the individual’s job separation occurred when the criterion switched from weeks to hours (post-January 1997). As for the previous estimates, a set of seasonal, age and education dummies are also included, along with the hourly wage at the time of separation, household equivalent income (at the time of the ROE), and regional unemployment rate.

A second version of the model is estimated with additional variables added to take into account of youth and re-entrant parents. That is, we include dummy variables for separators who were less than 25 years old and whose youngest child was aged 6 to 10. These variables are interacted with the policy dummies to test whether or not youth and/or re-entrant parents have been differentially affected by the policy changes, compared to other NEREs (see Table 7). Finally, the model is estimated separately for men and women. The probit results are presented in Table 8.

With regard to the effect of the policy changes on the access to unemployment insurance benefits among NERE workers, the results indicate that the change in the eligibility criterion from 20 to 26 weeks decreased the probability of NEREs receiving benefits, after controlling for other worker characteristics. The results indicate that the WEEKS26 variable is negative and statistically significant (see Table 7, column 1). However, the change in the eligibility criterion from weeks to hours of employment is associated with an increase in the access to benefits; this is indicated by the positive and statistically significant HRSBSE variable (see Table 7, column 1). These two results for the entire population of NERE workers also hold for women and men separately. Notice that when the model is estimated for male and female workers separately, the WEEKS26 and HRSBASE variables are negative and positive, respectively, and statistically significant for both the male and female equations (see Table 8, columns 1 and 3).

Apart from the policy changes, we also consider some of the other determinants of the probability of receiving benefits. For NERE workers, there is a lower probability of receiving unemployment insurance benefits associated with workers aged 15-24 years (compared to workers 35-44 years), single workers (compared to married workers), workers with children less than 18 years of age (compared to workers with no children less than 18 years), having an hourly wage rate of less than $10.00 (compared to $10.00- 14.99), and having a university degree (compared to a high school diploma). These results are suggested since the coefficients on each of these dummy variables (AGE15-24, SINGLE, YCHILT18, WAGELT7, WAGE710, and UNIV) are negative and statistically significant. Notice that the results for young workers, single marital status, presence of children less than 18 years of age, hourly wage rate, and education, confirm the results indicated by the descriptive analysis of incidence presented in Table 5b. With respect to education, notice that having less than a high school diploma is associated with a higher probability of receiving unemployment insurance benefits, as the LTHIGH variable is positive and statistically significant. Finally, being a new entrant (as opposed to a re-entrant worker) is associated with a lower access to unemployment insurance benefits; this result is indicated by a negative and statistically significant coefficient on the NEW variable.

But, are there any differential effects on our special interest groups (ie., youth and re-entrant mothers)? Results shown in Table 7 (column 2) indicate that both groups were particularly disadvantaged by the increase in the minimum weeks required for NEREs to qualify for benefits, with the impact not softened at all for re-entrant mothers by the switch from weeks to hours and the impact only partially softened for youth. Notice that the impact of the policy change was presumably primarily experienced by these two special interest groups because once we allow for differential effects for them, the WEEKS26 variable is no longer statistically significant for the NEREs overall.

The final set of regression provides estimates for NERE men and NERE women separately (Table 8). There are some differences between the results for the population of all NEREs and for men and women separately when we consider the various determinants of the probability of receiving benefits. For example, for men, the presence of children less than 18 years old is not associated with a lower probability of access to benefits (compared to men with no children); notice that the YCHLT18 variable is not significant. However, for NERE men, the presence of children 6-10 years old is associated with a lower probability of access to benefits since the YCH610 variable is negative and significant. For women, the presence of children less than 18 years old is associated with a lower probability of receiving benefits, as is the presence of children aged 6-10 years; notice that the variables YCHLT18 and YCH610 are negative and significant (Table 8).

In terms of the policy effects, women with children aged 6 to 10 experience a major reduction in access to benefits as a result of the increase in the minimum weeks, with no off-setting increase in access resulting from the change to an hours-based system. This large reduction in access to benefits is not evident for men with children aged 6 to 10.

Footnotes

⁹	There is no significant difference in the proportion of claimants receiving benefits in the two periods. An interesting research question is thus the impact of the program changes on the probability of filing a claim.
¹⁰	As discussed earlier, even after January 1997, claimants with working time during 1996 were granted 35 hours per week for each insured week of employment. Thus, we do not yet have data available which fully reflect the new system. However, Table 4 adds a fourth column, not carried through the remainder of the paper, which looks only at separations from July 1997 to December 1997 which would minimize this “transition problem”. Results are very similar, which we find reassuring.
¹¹	Since the increased eligibility condition for NERE workers was still in effect after January 1, 1997 (though re-expressed as 910=26X35 hours rather than as 26 weeks), all observations with job separations after January 1997 have WEEKS26=1 and HRSBASE=1. Observations from cohorts 5 and 6 with separations between July and December of 1996 have only WEEKS26=1.
¹²	We tried policy dummy interactions with this variable which were not statistically significant, so we do not include them here.
¹³	Household income is adjusted for the economies of scale available to those who live together using the Organization for Economic Cooperation and Development (OECD) equivalence scales.
¹⁴	The provincial unemployment rate varies both by province and year, partially controlling for changing economic circumstances over the period of study.
¹⁵	Since new entrants are both NERE and NEW, the two coefficients must be added to get the total effect for a new entrant.

Last Modified: 2000-11-15	Important Notices