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

• Purpose
• Background
• Methodology and Data Sources
• Key Findings
• Conclusions
• Biographical notes

Purpose

This study focuses on the behavioural changes of individuals during the switch from unemployment insurance (UI) to employment insurance (EI). Specifically, it investigates how the employment insurance (EI) legislation has affected individuals' unemployment durations and claimants' benefit durations (For further information on this area of investigation, see the study in this series by Stephen Jones, entitled "EI Impacts on Unemployment Durations and Benefit Receipt.").

Background

The Employment Insurance Act, which came into force in June 1996, was arguably the most fundamental restructuring of the UI program since the 1971 Unemployment Insurance Act. The scope of the revamping of the old UI system can be seen in the switch from the weeks-based to an hours-based system; the introduction of the intensity rule for repeat users; and some fine-tuning of the entrance requirements for new and re-entrants (NEREs), benefit levels, lengths of claim, repayment of benefits (the clawback provision), maximum insurable earnings, etc.

Previous legislative changes to UI have been shown to have an impact on workers' behaviour. If past experience is any indication, then we should expect Canadian workers to adjust their behaviour under the EI environment.

Methodology and Data Sources

The empirical investigation is based on the Canadian Out of Employment Panel (COEP) survey. The COEP survey consists of individuals randomly selected from Records of Employment (ROEs). The survey information is also linked to HRDC administrative data to yield a more complete data set.

The sample consists of nine consecutive cohorts. Each cohort includes individuals who separated from a job in a particular quarter. Thus the first cohort consists of people who separated from a job in July-September 1995, the second cohort consists of those who separated from a job in October-December 1995, and so on, through to the ninth cohort that consists of those who separated from a job in July-September 1999.

The COEP includes detailed information on a respondent's dates in and out of employment, reasons for breaks in employment, job-search activities, demographics of the respondent and his or her household, etc. Each survey uses a similar sampling methodology and identical survey instruments to ensure that the empirical findings would not be flawed with biased results originating from the data.

Since the changes brought in by the new EI legislation were phased in gradually, timing becomes a critical issue in the study's design. Specifically, the authors subdivide the post-EI period into two phases:

• Phase I: June 30, 1996 to December 31, 1996 — Implementation of the first set of EI changes
• Phase II: January 1, 1997 to December 31, 1997 — Implementation of the second series of changes in insurance parameters (The latest COEP interview available for this investigation did not go beyond the end of 1997.)

With these definitions, the authors might then estimate the effects of phases I and II according to the data reference periods. For example, the effects of phase I may be estimated by the Q95-4 (4th quarter, 1995) and Q96-4 (4th quarter, 1996) data, and the effects of phase II can be obtained by analyzing the Q96-4 and Q97-4 data. The total impact of EI can be calculated either from the phase I and II results or directly by using the Q95-4 and Q97-4 data. This methodology is frequently used to evaluate non-experimental programs.

In comparing post-EI data with pre-EI data, the authors essentially used the comparison group method of evaluation; however, recognizing the limitation of such a methodology, they augmented it with a variety of analytical tools. These econometric estimates, which controlled the different characteristics of job separators over time and the changes in the economic environments in the pre- and post-EI periods, provided the authors with the empirical evidence to reach their final conclusion.

Key Findings

The results of this study show that the EI legislative changes have shortened unemployment durations, and, to a lesser extent, benefit durations. Using the raw data, the authors calculated the probability of staying in an unemployment spell for the pre-EI and post-EI periods separately for all job separators — women, men, seasonal workers, nonseasonal workers, young adults (aged 25 or under), adults (age 25 and over), part-time workers, and full-time workers. The results for the pre-EI and post-EI periods for each group are plotted on the same graph. The most revealing information from this graphical analysis is that the post-EI curves lie everywhere below the pre-EI curves. Moreover, with the exception of the curve for young adults, the differences between the pre- and post-EI curves are statistically significant in every case. Thus the data show that the average duration of unemployment was shorter in the post-EI period than in the pre-EI period. Numerically the difference is about 1.5 weeks. To a lesser extent, these patterns are also evident for benefit durations.

The statistical technique on which these results are based does not account for possible differences in labour force characteristics in the pre- and post-EI periods and changes in the economic environments over time. To see whether or not these factors could have altered the conclusions, the authors estimate many econometric equations, and these results are entirely consistent with the previous ones.

The authors also show that under EI, individual groups did not behave identically. In particular, the study finds the following noteworthy differences:

• Men's unemployment spells were considerably shorter during phase I but not in phase II. Women's unemployment spells were slightly less sensitive to the legislative changes. Nevertheless, compared with the pre-EI period, women's unemployment durations were shorter in phase II, even after controlling for socio-economic factors over the pre- and post-EI periods.
• Overall, the EI legislative changes have not had a noticeable impact on the unemployment durations of young adults (aged 25 or less). Adults, in contrast, have been more sensitive to the legislation. During the 4th quarter of 1997, their unemployment spells tended to end earlier than did those of adults in the 4th quarter of 1995.
• EI legislative changes induced seasonal workers to leave their unemployment spells earlier than they would have in the absence of EI during phase I, but this effect did not persist in phase II.



Conclusions

In general, the econometric results indicate that the new EI legislation has considerably shortened unemployment spells and, to a lesser extent, benefit durations. As expected, various demographic groups have reacted differently to the new EI provisions.

These findings on the impact of EI on unemployment durations are consistent with Jones's findings. With respect to EI on benefit durations, the conclusions of the two studies are slightly different. The authors of this study agree with Jones that benefit durations in the pre-EI and post-EI periods are basically the same; however, after controlling for differences in labour force characteristics and economic climates, this study finds that the average benefit duration was slightly shorter in the post-EI period than in the pre-EI period. This minor discrepancy is probably due to the small differences in the econometric specifications of the two studies.

The data on unemployment spells also revealed that a number of individuals somehow managed to postpone their exit to some extent in order to benefit from the old UI rules and avoid benefit reduction if and when they experienced another unemployment spell. (For further information on this topic, please consult the study by Pierre Fortin and Marc Van Audenrode entitled "The Impact of Experience Rating on Unemployed Workers").



Biographical notes

Guy Lacroix is currently Professor of Economics at Laval University. He specializes in labour economics and applied econometrics. His research focuses on the interaction of income security policies and individuals' behaviour on the labour market (in particular welfare and unemployment insurance programs) and on the effect of the tax system on the black market.

Marc Van Audenrode is currently Professor of Economics at Laval University. His research concentrates on the impacts of institutions and regulations on the labour market. He has published extensively in academic journals and has worked as a consultant for both provincial and federal governments.

---







Last Modified: 2000-03-20	Important Notices