EI Impacts on Unemployment Durations and Benefit Receipt - October 2000

EI Impacts on Unemployment Durations and Benefit Receipt

View Whole Report

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

The 1996 Employment Insurance (EI) Act was one of the most fundamental reforms of the Unemployment Insurance (UI) program. The study provides an initial assessment of the switch from UI to EI during the early period of the EI system. It focuses on two issues: the impact of EI on the unemployment durations of individuals who experienced job separations, and the benefit durations of UI/EI benefit recipients.

Some of the changes brought in by the Employment Insurance Act came into effect on July 1, 1996 (e.g., the tighter eligibility rule for new entrants and re-entrants) and some came into effect in January, 1997 (e.g., the switch from a weeks-based system to an hours-based system). Several of the new provisions will take a number of years to become fully operational (e.g., the reduction of replacement rates for repeat users of the EI system). Changes were also made to benefit levels, the length of claims, the repayment of benefits (i.e., the clawback provision), and the maximum insurable earnings, among others.

Such drastic changes in the policy parameters are expected to bring about behavioural responses in terms of workers' job-search activities and firms' hiring decisions. The objective of this study is to assess these behavioural responses by estimating the effects of the overall EI legislative changes on individuals' unemployment durations and claimants' benefit durations.

The study by Guy Lacroix and Marc Van Audenrode, entitled "An Assessment of Various Components of C-12 on the Duration of Unemployment Spells," addresses similar issues and, overall, the findings of that study and this one are consistent with each other. What minor discrepancies there are probably occur because the econometric specifications of the two studies are not identical.

The study is primarily based on the Cox partial likelihood duration model, which is widely accepted as one of the appropriate methods for analysing duration data. This method takes account of durations that do not end at the time of the survey, the rate at which people leave unemployment spells, and how benefit receipts vary with duration lengths.

Two models are estimated. The first compares individuals in the same quarter in different calendar years, before and after the EI reform. The second pools all the survey data together and controls for the confounding effects of seasons by econometric treatments (dummy variables) in the estimated equations. While the first model is a more flexible way to separate the seasonal effects from EI impacts, the larger sample size in the second model allows the authors to assess EI impacts by demographic group.

The robustness of the results was tested by re-estimating alternative duration models of different functional forms. In general, the results were robust and the findings could not be significantly changed by different specifications.

The study employed 1996 Canadian Out of Employment Panel (COEP) survey data along with UI/EI administrative records. In the COEP, all surveyed participants experienced at least one job separation or break/change in employment between July 1995 and December 1997. There were 10 cohorts in the survey. Cohorts 1 to 4 consisted of those who separated from their jobs under the UI system (from July 1995 to June 1996), and cohorts 5 to 10 consisted of those who separated from their jobs under the EI system (from July 1996 to December 1997). These data enabled the author to use a quasi-experimental design and econometric methods to assess the EI impacts.

Change in unemployment durations

The shift from UI to EI resulted in a clear but small reduction in the duration of unemployment among those separated from their jobs. The small effects suggested that there was no dramatic shift in the determinants of the duration of unemployment as a result of the move to EI.

The behavioural effects were quite different for the various individual demographic groups. Reductions in unemployment spells were observed for men, workers aged 25 years and over, and for those who separated from their jobs due to "shortage of work or other reasons." This effect was not noticeable for women, youth, and for workers who voluntarily quit or were dismissed from their jobs. In other words, the unemployment durations for these individuals remained more or less unchanged.

Change in benefit durations

The author found that the transition from UI to EI did not result in changes in benefit durations. This lack of convincing effects applied to all individuals who experienced a job separation and the different demographic groups of job separators. The absence of significant results for all job separators was thus not a statistical artefact resulting from opposite effects among individual demographic groups.

The study by Lacroix and Van Audenrode reported that, compared to the pre-EI period, EI had shortened the average benefit duration slightly in the post-EI period. This minor discrepancy was probably due to the slightly different econometric specifications and treatments of the two studies.

The move to EI resulted in a significant but small decrease in the expected unemployment durations. This effect was noticeable for men, workers aged 25 or above, and workers who left their jobs because of "shortage of work or other reasons." There was very little evidence that benefit receipt duration changed with the switch to EI.

Stephen Jones is currently Professor of Economics at McMaster University. His research covers many areas of labour economics, with particular focus on issues related to unemployment. He recently completed a three-year term as co-editor of the Canadian Journal of Economics. Currently he is directing the Labour Market Institutions and Unemployment project with the Canadian International Labour Network. His past consultative work includes a major study on unemployment persistence and hysteresis for the Economic Council of Canada; the initiation, design and implementation of the Canadian Out of Employment Panel data set funded by HRDC; and various studies of UI legislative changes (Bills C-113 and C-17) for HRDC using the COEP93 and COEP95 data sets linked to administrative files.