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

The conclusions from this examination can be broken down into three main areas.

1. The new system changes both eligibility and entitlement.

The switch to the Employment Insurance (EI) system led to a net increase in eligibility. Very small numbers of workers who would have qualified for benefits under Unemployment Insurance (UI) are ineligible under EI. These tend to be female, part-time/part-year service workers. Since this tends to be a vulnerable group, some attention should be paid to them, even though it is a very small group.

The switch to the EI system leads to an increase in entitlement, especially for seasonal workers who work 10 to 20 week jobs (the range of weeks corresponding to entrance requirements in the old UI system). This is true because seasonal workers tend to work long hours per week and because the EI eligibility and entitlement table values (reported in hours) are just the old UI table values (reported in weeks) multiplied by 35. Thus, a given number of weeks of work generates more weeks of entitlement under EI than under the old UI system if the usual hours per week on the job is over 35.

These entitlement and eligibility effects are dwarfed by the negative eligibility and entitlement effects generated by the 1994 UI changes.

In spite of net overall gains in entitlement, both the group that experiences an increase in entitlement and the group that has a decrease in entitlement in response to the shift to EI are sizeable. Approximately 8 percent of workers have reductions of over 5 weeks of entitlement in the switch to an hours-based system versus 11 percent who have increases of more than 5 weeks.

The groups that experience increases and decreases in entitlement are quite different. Not surprisingly, given that the old weeks-based entitlement and eligibility requirements are converted to the EI hours-based requirements by multiplying by 35, the group that has declines in entitlement tends to be individuals working part-time (less than 35 but more than 15 usual hours per week) and the group that experiences increases in entitlement tends to be individuals working long hours (more than 35 hours per week). In addition, apart from the reduction in maximum entitlement from 50 to 45, a change in entitlement resulting from the system change will only be observable if the individual worked less than the amount needed to qualify for maximum benefits in 1996. Thus, both groups will tend to be part-year workers. The group experiencing increases in entitlement tend to be male seasonal workers, while the group experiencing declines in entitlement tend to be female part-year, part-time workers. Both groups have average wages above the average for seasonal workers as a whole, but both also have significant numbers of low wage workers. For the group with entitlement reductions in particular, there appear to be two groups: better paid, unionized workers in the public sector and lower paid, non-unionized workers in the service sector.

Our overall conclusion in terms of eligibility and entitlement is that the switch to an hours-based system had almost no impact on eligibility but did increase the number of weeks of benefits workers were entitled to, in general. A shift from a weeks-based to an hours-based system redistributes weeks of entitlement toward long hour, predominantly male, seasonal workers and away from part-year/part-time, predominantly female workers. The average wages for all those affected by the shift to EI are above the average wage for seasonal workers as a whole.

2. The examination of the distribution of weeks worked in an employment spell leads to several clear conclusions.

There is evidence in the pre-1997 data of UI effects on the weeks of work distribution but these effects are generally not large. In the pre-1997 period when the entrance requirement is weeks-based, duration model estimates indicate a clear entrance requirement (ER) effect for seasonal workers but not for non-seasonal workers. Even for seasonal workers, however, the effect is not huge. Our estimates indicate that just under 1 in every 100 seasonal workers whose jobs last at least as long as the ER experience a job termination at the ER point because it is the entrance requirement and they have just qualified for UI. Slightly smaller effects are found for incentives induced by seasonal workers working just enough weeks to qualify for the maximum possible benefits they can receive from UI before the start of the next season. Together, these effects imply that a 2-week reduction in the entrance requirement induces only a 0.3 week reduction in average employment spell duration for seasonal workers.

In the post-1997 period, there is clear evidence of adjustment to the new system. The hazard rate rises substantially when employment spells pass the ER point and there is a spike at the MAXYR point. Since the MAXYR point changes substantially for many workers under the EI versus the UI system, this suggests some quite sophisticated tailoring of employment spells to the new EI system.

Estimates using post-EI change data indicate an increase in responsiveness to entrance requirement changes among seasonal workers. A 2-week reduction in the entrance requirement generates a 3/4 of a week reduction in average employment spell durations for seasonal workers in 1997. This is quite a large reaction implying that the shift to an hours-based system has led to noteworthy reductions in job lengths among seasonal workers.

The fact that there still appears to be a spike in the 1997 hazard at the number of weeks that would have been the entrance requirement under the old system indicates that a full adjustment to the new hours-based system may not yet have taken place.

There is little evidence of any adjustment to the UI or EI systems by non-seasonal workers in either year. A 2-week reduction in the entrance requirement does not change the average duration of employment spells for non-seasonal workers.

The hazard rate is higher in the 12 to 15 week range in 1997 relative to 1996 for seasonal workers. As a result, the average duration of a seasonal employment spell is 1.5 weeks shorter in 1997 than 1996. These differences between the two years appear to be due both to larger reactions to UI/EI parameters in 1997 and a higher probability of job durations falling in this week range in 1997 even in the absence of any measured UI/EI effects. This might be a reaction to the new EI system but more observation is required to be certain of any such conclusion.

Our overall conclusion with respect to impacts on employment spell durations is that the shift to the hours-based system has had noticeable impacts in terms of shortening seasonal employment spells. This is plausible since seasonal workers typically work long hours per week and, as a result, can acquire given levels of EI entitlement in fewer weeks under the new system. However, this effect should be kept in perspective. Non-seasonal employment spells, which make up the great majority of all employment, show no patterns of adaptation to the UI/EI systems or changes with the introduction of the EI system.

3. There is virtually no change in the usual hours per week distribution between 1996 and 1997.

In particular, the distribution of usual hours per week for seasonal workers in high unemployment regions, the group that appears to react most to UI incentives, does not change. Thus, while there does appear to be some reaction to the new system in terms of duration of jobs measured in weeks, there is no apparent reaction in terms of adjusting hours per week.