Evaluation of the Pilot Project to Extend Employment Insurance Benefits by Five Weeks: 2004-2009, March 2010

List of Abbreviations

• COEP - Canadian Out-of-Employment-Panel
• EI - Employment Insurance
• GDP- Gross Domestic Product
• HRSDC - Human Resources and Skills Development Canada
• MAR - Monitoring and Assessment Report
• NERE- New Entrant/Re-entrant to the Labour Force
• ROE - Record of Employment
• SV - Status Vector
• VER - Variable Entrance Requirements
• WWOC - Increase in Allowance Earnings

Executive Summary

A pilot project was initiated in June 2004 that extended Employment Insurance (EI) entitlements for individuals claiming regular benefits by up to five additional weeks and up to the regular maximum of 45 weeks. Research has shown that there is a segment of the seasonal claimant population who experience gaps in income (referred to as seasonal gappers), during which they do not receive EI benefits or employment income. The goals of the pilot were to test the extent to which the extension of benefit weeks reduces the number of seasonal claimants experiencing an income gap, whether any unintended behavioural effects have been observed, and estimate the cost of the benefit extension. The pilot project ran from June 6, 2004 to February 28, 2009 in selected high unemployment economic regions. Initial estimates forecasted that the extra cost would be about $100 million per year, with 100,000 seasonal workers benefiting annually.

Overall, the pilot project was effective in its primary goal of reducing the number of claimants who are seasonal gappers. Prior to the pilot, approximately 1.4 percent of claimants in the pilot regions were seasonal gappers, while in the first year after the pilot's implement­ation this figure fell to 0.7 percent. Estimates show that 31.1 percent of all claimants in the pilot regions received at least one additional week of benefit, but comparing this to the fact that less than 2.0 percent of claimants are seasonal gappers suggests that the pilot project was not effective at only targeting the seasonal gapper population.

Estimating the true costs of the pilot has proved to be problematic as the pilot not only provided additional weeks of EI benefits to claimants, but also changed the behaviour of many of those who received benefits. It was found that claim spells were longer for a significant proportion of claimants, and employment spells to qualify for EI were shorter. These changes in employment spell duration could be a function of both employee and employer interactions that occur as a result of the pilot project. The orders of magnitude of these effects were not found to be large in the short term, but claim duration increased while weeks and hours worked decreased as the pilot project progressed. The fact that some claimants used additional weeks of benefits might be associated with them using this time to find a better job match. It is also interesting to note that while many claimants tended to work fewer hours during the EI qualifying period as a result of the pilot, gappers observed a larger reduction in work hours than non-gappers over the first three years.

These behavioural changes are significant enough to increase the costs over and above the costs that had been estimated before the implementation of the pilot. Although the project has been found to be successful in reducing the proportion of claimants who are seasonal gappers, the chosen instrument has proven to be an expensive means to achieve this end. Assuming no behavioural change, the estimates of actual costs paid out by the program are between $160.8 million and $192.8 million per year over the first three years, while incorporating a behavioural change inflates the costs by even more, as the annual cost in each of the first three years of the pilot is estimated to be between $257.8 million and $283.5 million. This does not include the opportunity cost to the economy of having workers work fewer hours prior to being on claim. Estimates suggest that the pilot could have led to a reduction of more than 5 million work hours in the first year of the pilot and almost 11 million work hours in the second year of the pilot.

The pilot was successful in demonstrating that the rules implemented did reduce the number of seasonal workers who experienced gaps in earnings. It also showed that the cost of implementing the rules is fairly high primarily because the proposed instrument provides a significant amount of benefits to non-seasonal workers. Additionally, the pilot leads to a significant reduction in work hours inducing a higher cost to the economy as a whole.

Management Response

Evaluation Observations

Among seasonal gappers, the group targeted by the pilot, the evaluation reveals that the proportion of claims that were made by gappers declined by nearly 50%. For the other 50% of gappers, the additional weeks of benefits helped to reduce the number of weeks without any income.

The evaluation results outline the number of additional weeks of benefits received by claimants of different profiles. As expected, seasonal gappers used nearly all of the additional five weeks offered under the pilot project. This result confirms that most of them were well below the 45-week entitlement maximum before the pilot project and could therefore benefit from the 5 extra weeks of benefits.

Although the measure targeted a specific claimant population, the seasonal gappers, other claimants were able to benefit from the additional weeks of benefits since the pilot project was in effect in all of the high unemployment EI regions and not limited to seasonal gappers. The evaluation results found that gappers worked fewer weeks and a reduced numbers of hours prior to establishing a claim over the first three years of the pilot project.  In addition, the results show that claimants who were not gappers received additional weeks of benefits in both the first and second year of the pilot project.

In general, the direction and magnitude of the impact that the pilot project had on claimants in the targeted regions are consistent with results from the analysis conducted by the program area. The use of additional weeks of benefits could sometimes lead to positive outcomes, as pointed out in the evaluation, if the additional time allows claimants to broaden their job search in order to find a job that matches their skills.

The evaluation shows that of the total amount paid under the pilot project only 2.3% was received by seasonal gappers, while non-seasonal gappers received 10.7% and non-gappers 87.0%. These findings are again consistent with results observed by the program area.

The higher than originally estimated cost is mainly explained by the fact that behavioural changes were not considered in the estimated, initial cost and the pilot project was targeted to high unemployment regions and not specifically to seasonal gappers.

Considerations

Pilot projects are intended to allow the Government to test the labour market impacts of new approaches before permanent changes to the EI program are considered. The evaluation results of the five-week pilot project provide a good example of why the pilot project provision is necessary for some program changes. The intention of the pilot was to address the seasonal gap faced by workers who experience a period without employment income or EI benefits.

The evaluation results show that the objective of reducing the number of seasonal gappers or the length of the income gap was met. They also indicate that claimants, gappers or non-gappers, worked fewer hours and benefited from the additional weeks of benefits, which increased the cost of the pilot project. Program analyses conducted on these pilot projects also concluded that a significant portion of the cost was associated with additional benefits paid to non-gappers. Overall, the evaluation results have demonstrated that there were non-intended behavioural changes in response to the pilot project.

These results are important and will be taken into account when designing future pilot projects. They provide a clear indication that testing program changes through pilot projects is important to assess all impacts, including behavioural effects, and finding a means of better targeting any future measure will be important.

1. Introduction

This evaluation assesses the impact of Pilot Projects #6 (Pilot project on increased weeks of Employment Insurance (EI) benefits) and #10 (The Extended Employment Insurance (EI) Pilot Project), which were implemented in June 2004 and June 2006 respectively. These two pilot projects, collectively known as the 5-week seasonal pilot project, extend regular EI benefits by up to 5 weeks for claimants living in high unemployment regions. ^{Footnote 1} This pilot project makes use of a special feature of the EI system which allows temporary changes to be made to the EI system in order to test for possible improvements to the system. This pilot project was implemented on June 6, 2004 and ended on February 28, 2009 in selected EI economic regions, covering about 50 percent of all regions across Canada. ^{Footnote 2}

The decision to extend benefit weeks stemmed from a 2004 report produced by the government's Seasonal Task Force. ^{Footnote 3} The concern was motivated by observing a significant number of unemployed seasonal workers who experienced a stretch of zero-income weeks after exhausting their EI benefits and before starting their next job. These seasonal workers are sometimes referred to as "gappers", as there is a black hole ^{Footnote 4} in the flow of income from either work or EI benefits.

In 2005, a number of other temporary measures were introduced into the EI system, in addition to the 5-week seasonal pilot project. The other three pilot projects for EI beneficiaries with regular benefits are the Best 14 Weeks, the Increased Access to EI and the Increase in Allowable Earnings. ^{Footnote 5} These three pilot projects were implemented between October and December 2005. They took place in 23 of the 24 high unemployment EI regions that were included in the original version of the 5-week seasonal pilot project. ^{Footnote 6}

This report will give evidence on three basic themes relevant to the evaluation of the 5-week seasonal pilot project. It will focus on the ability of the pilot project to effectively target its primary subject, seasonal gappers. It will also discuss the pilot project's influence in changing EI claimant's behaviour, with respect to how long a claimant stays on EI, the number of weeks and hours worked during the EI qualifying period, and job search behaviour while unemployed. The final theme will be the cost of the pilot project, including the cost of the additional weeks of benefit and a discussion of the economic costs associated with changes in workers' behaviour. Additionally, there will be a brief discussion of employer attitudes in regards to the pilot project.

The paper discusses these three themes throughout the following sections: Section 2 provides an overview of the 5-week seasonal pilot project and Section 3 provides a discussion of the methodology that was used in evaluating this project. Section 4 provides the analysis of the pilot project's rationale while Section 5 discusses the issues relating to targeting and the achievement of the pilot's objectives. Section 6.0 reports on the behavioural impacts that are attributable to the implementation of the pilot project. Section 7.0 presents the response of seasonal employers in the wake of employees being able to stay on EI for up to five weeks longer. Section 8.0 presents a discussion of the costs associated with the pilot and Section 9.0 contains the conclusions of the evaluation report.

2. Overview of the 5-Week Seasonal Pilot Project

2.2 Definitions

The study of this pilot project requires the use of terms and concepts that are not immediately obvious. This section defines some of the key terminology that is used throughout the paper.

This report studies the behavioural impacts for different groups of claimants, namely:

All (Pure Regular Claimants):
All claimants who only received regular benefits during the course of their claim.  This includes gappers and non-gappers.

First-Time Claimants: 
Claimants who had no regular claims prior to their current claim.

Occasional Claimants: 
Claimants who had one or two claims in the five years prior to their current claim.

Frequent Claimants: 
Claimants who had three or more regular claims in the five years prior to their current claim.

Seasonal Claimants:
Claimants who had three or more regular claims in the five years prior to their current claim, all starting in the same 17-week calendar period ^{Footnote 9} as their current claim.

Gappers:
Individuals who took more than one week but less than fifteen weeks to find their next job after their previous finished claim. This includes individuals who were first-time claimants in the previous completed claim.

Seasonal Gappers:
Gappers who had three or more regular claims in the five years prior to their current claim, all starting in the same 17-week calendar period as their current claim. 

Non-Seasonal Gappers:
Gappers who had less than three regular claims in the five years prior to their current claim and also those gappers who had three or more regular claims in the five years prior to their current claim but whose claims did not start in the same 17-week calendar period as their current claim. This includes individuals who were first-time claimants in the previous completed claim.

Non-Gappers:
Individuals who did not exhaust their current claim as well as those who took one week or less to find their next job after their previous completed claim.

Seasonal Non-Gappers:
Non-gappers who had three or more regular claims in the five years prior to their current claim, all starting in the same 17-week calendar period as their current claim.

Non-Seasonal Non-Gappers:
Non-gappers who had less than three regular claims in the five years prior to their current claim and also those non-gappers who had three or more regular claims in the five years prior to their current claim but whose claim did not start in the same 17-week calendar period as their current claim. This includes first-time, occasional and frequent claimants.

Exhaustees:
Claimants who have used up all their weeks of EI benefit entitlement.

Seasonal Exhaustees: 
Seasonal claimants who have used up all their weeks of EI benefit entitlement.

Chart 1 below shows a graphical representation of what constitutes a gapper versus a non-gapper.

It is important to note that the classification of individuals is based on their previous claim history, use of benefit entitlement and subsequent employment spell. This means for example that an individual is not a gapper until they have finished their claim and experienced a gap in earnings. With maximum weekly benefits increasing by up to five weeks in the pilot regions, it is possible that someone who would have been a gapper under normal EI rules is not a gapper under the pilot rules. Similarly, someone who would have experienced an income gap of 16 weeks and would not have been considered a gapper under normal EI rules may only have an income gap of 11 weeks under the pilot project and is now considered a gapper in the pilot period. As such, care needs to be taken when interpreting results attributed to the different groupings of claimants between the pre-pilot and pilot periods, since these groups will not be made up of the exact same people. ^{Footnote 10}

2.3 Rationale for Implementing a Pilot Project

Through Part V, Regulation 109 of the Employment Insurance Act, Human Resources and Skills Development Canada (HRSDC) is allowed to implement temporary changes to EI without passing any new piece of legislation. This way, the department has some freedom to test the impact of changes to the EI system, before inscribing the new rules permanently into legislation. This feature has been used occasionally over the years. A recent example is the Small Weeks provision in the Employment Insurance Act which was tested and evaluated twice before the Small Weeks Pilot Project was formally adopted into legislation.

Adding up to five weeks to EI entitlement could have significant cost implications. To illustrate, if there are one million regular claimants across the country and 20 percent of them use all their benefits (exhaust their benefits) then the cost of a one week extension of the benefits at $300 per week would be $60 million. The 5-week seasonal pilot project extended entitlement by up to five weeks, suggesting the total impact on cost could be substantial.

Similarly, other potential complications can arise from the extension of entitlements to EI. It is possible that some claimants may change their behaviour. For example, claimants may remain unemployed for a longer period of time if the duration of benefits increases. Some clients may work less before starting a claim, since it takes fewer hours of work to qualify for the same number of benefit weeks. These two potential behavioural changes imply that it is extremely difficult to know, prior to implementation, how much the changes to the EI system will affect work behaviour. This provides further justification for the need to evaluate the impact of potential changes to the system.

2.4 Details of the Program Change

The EI system is inherently complex. A brief summary of the rules as they relate to entitlement weeks is contained in Tables D1 and D2 in Annex D for the convenience of the reader. More complete descriptions of the EI system are available in the technical reports cited in Annex C.

Each claimant is allotted a certain maximum number of weeks of benefits, depending on the local unemployment rate and the number of hours worked during the EI qualifying period. The minimum number of benefit weeks a claimant can be entitled to is 14 weeks in low unemployment regions, and the maximum is 45 weeks in high unemployment regions. The pilot project extended EI entitlement by up to five weeks in specific EI economic regions. However, claimants could still only receive the standard maximum of 45 weeks of benefits. Thus, if a given individual was eligible to receive 39 weeks before the pilot, they would be entitled to 44 weeks during the pilot. However, if they had been entitled to 41 weeks before the pilot, they would only be eligible to receive an additional four weeks of benefits. The chosen EI regions were selected based on whether the region reached a 10 percent unemployment rate during the six-month window prior to May 2004.

2.4.1 History and Timing

The pilot was implemented in two phases. Pilot Project #6 was approved for implement­ation in May 2004. It allowed for the number of entitlement weeks to be increased during a two-year period. As this two-year period came to an end, it was becoming clear that the data had not yet matured enough to determine the effectiveness of the pilot. As a result, when the first pilot ended, a new pilot (Pilot Project #10), similar in structure, was initiated for another year-and-a-half to confirm the validity of the results. The pilot ended on February 28, 2009. ^{Footnote 11}

The initial two-year pilot period appears lengthy, but was not long enough to collect the data required to adequately answer key questions. For example, it is important to determine which claimants are seasonal gappers. In order to do this, a seasonal worker must be observed ending a job, going on EI and then starting a second job. The start date of the second job is compared to the end date of the receipt of EI benefits to determine if a gap exists. However, information about this second job is not known until the job has ended and a Record of Employment (ROE) is issued. Furthermore, the timing of the data collection is such that the information about the second job ROE will not appear in HRSDC databanks until several months after the job has ended.

Thus, towards the end of the second year of Pilot Project #6, only a partial sample was available. From this sample, it was apparent that there were interesting findings with regards to the targeting of the program and changes in behaviour. Still, since the sample was partial, it was decided that more data had to be collected and tested to increase the degree of certainty. On June 11, 2006 the pilot was extended in the form of Pilot Project #10, which had its regulations go into effect on June 23, 2006, ^{Footnote 12} and extended the pilot to February 28, 2009.

2.4.2 Affected Regions ^{Footnote 13}

The pilot regions for Pilot Project #6 were those that had more than 10 percent unemployment for at least one month in the six-month period before May 2004. ^{Footnote 14} This list was updated with the second pilot (Pilot Project #10) to exclude three EI regions where the economic situation substantially improved and unemployment rates were consistently below 8 percent.

One means of measuring program impacts is to compare the behaviour of clients in a given region with those in similar regions before and after the change to the system. Since the pilot project is taking place in some regions and not others, it is possible to perform this type of analysis by comparing claimant behaviour in pilot regions to those in non-pilot regions. It should be noted that since the pilot is only taking place in high unemployment areas, the comparison between pilot and non-pilot regions is not ideal. However, statistical analysis and sensitivity testing will allow for reliable conclusions to be drawn.

3. Methodology

The approach in this study is multi-faceted using a number of data sources and several lines of evidence. Evidence is collected to look at the rationale of the pilot, the ability of the pilot to achieve its objective of eliminating the income gap faced by many seasonal workers, determine if there are any unintended effects encountered as a result of the pilot, and the cost of the pilot. The data sources include EI administrative data, survey data and expert interviews.

Since the identification of seasonal gappers in the data is challenging, the project also examined the impacts on seasonal claimants who exhaust their benefit weeks, but who do not qualify as seasonal gappers under the definitions described earlier. Roughly 30 percent of claimants use all their weeks of entitlement before the end of the 52-week claim period. ^{Footnote 15}

A series of studies have been conducted for this evaluation project, including quantitative analysis and qualitative analysis. Academic advisors were contracted to oversee the technical aspects of the evaluation project, primarily because there are complex data issues and the chosen methodology is fairly technical.

3.1 Multiple Lines of Evidence

A literature review was conducted to provide an understanding of the theoretical reasons for, and existing findings on, EI support of seasonal workers. A series of technical reports were prepared to determine the characteristics of gappers and the impact of the pilot project on the behaviour of claimants. The reports were produced internally as well as contracted out to academic consultants. An academic advisor was consulted during the development of the internal reports to ensure the methodology was correctly applied and the interpretations of the results were reasonable.

The rationale for implementing this pilot project is based on the assumption that those experiencing income gaps are undergoing considerable hardship. To aid in assessing this rationale, a literature review of the existing research on seasonal workers and a statistical review of pre-pilot seasonal claimants were conducted. The statistical review examined data available for the pre-pilot period to gain an understanding of the household, work and economic environment of seasonal workers.

Qualitative analysis is an important tool to help understand the impact of EI and the pilot project on employers. Consequently, two sets of key informant interviews were conducted with each of the following groups: employers, Service Canada officials and union representatives. One set of interviews occurred in the fall of 2005, and the second took place in the fall of 2006. These interviews allow the impact of the pilot projects to be viewed from the perspective of the employer, something that could not be done with quantitative analysis alone. These interviews give a sense of the extent to which employers are being affected by the extended benefit weeks and whether standard hiring and layoff practices had changed because of the pilot. With the two interviews approximately one year apart, changes in opinions and attitudes over time can be tracked as well.

While this evaluation looks at whether the objectives of the program are being achieved and the cost of the pilot project, there is further work that considers if seasonal gappers experience significant economic hardship (McDonald & Gray, 2006) as well as if there are any unintended behavioural changes occurring as a result of the pilot (HRSDC, 2008, Human Resources and Social Development Canada, March 2006, Stewart, 2007).

3.2 Data Sources

The internal and external statistical studies use HRSDC administrative data files along with some survey information. The primary database used is the EI Databank which contains the Status Vector (SV) and Records of Employment (COEP). These data sources contain detailed claim and employment information. The other data source is the set of surveys known as the Canadian Out-of-Employment-Panel (COEP) survey data, which is linked directly to the SV and the ROE databases. The COEP yields otherwise unattainable information on the household status of claimants, income information and additional employment information not found in the administrative data.

Some of the tables included in this report were taken from other reports (Tables A3 to A10) while the remainder were created solely for this report (Tables 1, A1, A2, A11 and A12). The tables created for this report use SV and ROE data and were put together using a 20 percent sample of individuals who experienced a claim between January 2001 and December 2007.

Identifying gappers in the administrative data proved to be a challenging task, since a fairly long period of time must pass before the necessary information appears in the administrative data files. To identify gappers in the post-pilot administrative data, claimants need to exhaust their benefits, experience an income gap, find alternative employment, work for some time, and then stop working for some reason so that an ROE can be issued. The implication of this is that not all gappers will be observed. Those individuals who exhaust their benefits but then obtain a permanent job that is never lost will never have another ROE issued, meaning it is impossible to know if and when they obtained new employment. This results in the gapper analysis only being able to focus on those with subsequent employment durations that are shorter in nature.

For this entire cycle to be complete, particularly when examining claimants who initiate their claims during each season of the year, two full years of data are needed to determine if an individual is a gapper. A detailed discussion of the timelines needed for complete data is included in Table D4.

4. Analysis of the Program Rationale

Applying separate rules for seasonal workers assumes that they are unique in their difficulties in establishing a regular income stream. However, a literature review has shown that seasonal workers are heterogeneously distributed across the workforce and that seasonal work is not necessarily associated with economic hardship. ^{Footnote 16} The literature review revealed that seasonal work is not dominated by low-income households or households with only one earner. ^{Footnote 17} Further evidence exists to show that seasonal workers have about the same level of annual income as other workers who experience a job loss. ^{Footnote 18}   Furthermore, seasonal claimants are less likely to exhaust their claims and experience a gap than other earners, in particular, occasional claimants. In percentage terms, 29 percent of occasional and 22 percent of seasonal workers are likely to exhaust their claims, and 14 percent of occasional and 5 percent of seasonal workers are likely to experience a gap. ^{Footnote 19}

It has been suggested that seasonal gappers are as able as other claimants to smooth their consumption over periods of joblessness, meaning that these workers do suffer a large drop in their standard of living as the result of a job loss. ^{Footnote 20} They are less likely to stay unemployed over the course of the year than other workers—primarily because their work is seasonal. Overall, these findings would suggest that seasonal workers are not unique in their difficulties of establishing a regular income stream.

5. Targeting and Achievement of Stated Objective

The stated objective of the 5-week seasonal pilot project is to reduce, and preferably eliminate, the gap in earnings that seasonal workers experience after exhausting their EI benefits and before they find subsequent employment. With research showing that seasonal workers are heterogeneously spread over the workforce, ^{Footnote 21} it is important to make sure that this program reached its target audience.

The analysis attempts to look at the effects of the pilot immediately after implementation as well as later on when the pilot has been in place for a significant period of time. The data is split between seasonal gappers and other groups, such as non-gappers. This way, it is possible to compare to what extent the extra weeks are benefiting seasonal gappers relative to other claimant types.

5.1 Change in the Proportion of Seasonal Gappers and Exhaustees

The first quantitative issue of concern is whether there are fewer seasonal gappers since the start of the pilot project. In previous studies, seasonal gappers have been shown to be highly sensitive to changes in the policy environment. In fact, Bill C-21 (1990), which increased benefit duration, led to a decline in the percentage of seasonal workers experiencing an income gap from 43.9 percent in 1990 to 21.8 percent in 1991. After Bill C-17 (1994), EI entitlements were cut substantially, leading to this percentage increasing from 22.0 percent in 1993 to 34.1 percent in 1994. Finally, after the 1996 EI reform increased entitlement for those who worked more hours per week, the percentage of gappers in 1997 declined to 26.9 percent from 36.2 percent in 1996. ^{Footnote 22}

The pilot data shows that there has been a considerable decline in the incidence of seasonal gappers. From the sample of data collected, the proportion of claimants who were seasonal gappers in the June 2003-May 2004 months in the pilot regions was about 1.4 percent, and this dropped to 0.7 percent in June 2004-May 2005, a decrease of about 50 percent. The data for the year June 2005-May 2006 shows that the proportion of claimants who are seasonal gappers increased slightly over the June 2004-May 2005 year. ^{Footnote 23}

The decrease in the proportion of seasonal exhaustees was similar to the decline of seasonal gappers, at close to 40 percent from June 2003-May 2004 to June 2004-May 2005. ^{Footnote 24} In the following year, there was an increase in this proportion relative to the first post-pilot year, but still a decrease from the pre-pilot period of 32.4 percent. ^{Footnote 25}

5.2 Success in Targeting the Program

5.2.2 Seasonal Exhaustees

It is reasonable to broaden the group of interest to include all seasonal exhaustees, on the assumption that these individuals have the same difficulty finding suitable re-employment as gappers. The only difference between gappers and exhaustees, is that gappers are a subset of exhaustees who find new employment within a 14 week window after the end of their EI entitlement. Exhaustees on the other hand exhaust their EI benefits and then may or may not find new employment. Clearly seasonal exhaustees also benefit from this pilot project.

In general, Table A2 shows that the pattern found for seasonal gappers and non-gappers holds for seasonal exhaustees and non-exhaustees. About 87.8 percent of exhaustees and 97.8 percent of seasonal exhaustees benefited from the pilot project. On average, all exhaustees received an additional 4.1 weeks of benefits while seasonal exhaustees received an additional 4.8 weeks of benefits. ^{Footnote 31} As expected, those who do not exhaust their benefits use fewer additional weeks. Only 10.4 percent of all non-exhaustees and 12.5 percent of seasonal non-exhaustees received any extra weeks of benefit. ^{Footnote 32}

According to Table A2, while 74.2 percent of all exhaustees took the full five weeks of extra benefits, almost all of the seasonal exhaustees, 93.4 percent, took the full five weeks. Only 2.2 percent of all seasonal exhaustees took zero extra weeks of benefits. This would suggest that very few seasonal exhaustees were already at the maximum 45 week entitlement threshold without the pilot rules being in effect. The results suggest that 89.6 percent of all non-exhaustees and 87.6 percent of seasonal non-exhaustees did not benefit from the pilot project. The results suggest that 24.6 percent of pilot beneficiaries were non-exhaustees, while 8.5 percent were seasonal exhaustees. ^{Footnote 33}

5.2.3 Summary

Overall, the targeting of the project is adequate in that almost all seasonal gappers and seasonal exhaustees benefit from the program. While they received close to the full five weeks of extra benefits, the reason they did not receive the full five weeks is because some individuals would have been eligible for four or fewer extra weeks, since entitlement is still capped at the pre-pilot maximum of 45 weeks. However, there is a significant portion of non-gappers and non-exhaustees that were also taking extra weeks of benefits. While this seems strange on the surface, since non-gappers and non-exhaustees should not require any additional benefit weeks, the reason they used these extra benefits may be because they would have been gappers or exhaustees had the additional benefits not existed. It is impossible to tell whether the additional weeks on claim are a result of the individual choosing to stay on benefits longer or because they actually needed the extra entitlement to carry them over their entire unemployment spell. It is also important to note that the figures presented in this section do not account for behavioural changes caused by the pilot project. Because of this, it is possible that the changes in benefit weeks received are a result of other factors, such as a large increase in the unemployment rates in the pilot regions causing benefit weeks to increase even in the absence of the pilot project. Section 6 will use more sophisticated statistical techniques to address this issue.

6. Impacts Attributable to the Pilot Project

As with any program change, there will be outcomes, either intended or unintended, that inherently stem from the extension of entitlement weeks.  Care needs to be taken to be certain which of these impacts can be attributed to the pilot project.

This section reports on the impact of the pilot project on weeks of benefits received, as well as weeks of work in the EI qualifying period and insured hours for both gappers and exhaustees (those targeted by the program) and non-gappers and non-exhaustees (those not targeted by the program). Changes to weeks of work and insured hours are unintentional effects of the pilot project and it is important that these findings are reported in addition to the intended behavioural changes. It is important to note that behavioural changes related to weeks of work and hours of work are not simply the result of workers changing their behaviour, but also the result of employers being influenced by the implementation of the pilot project. The complex relationship between employees and employers makes it difficult, if not impossible, to separate the impact of the pilot project across all of these stakeholders.

It is difficult to isolate the impact of the 5-week seasonal pilot from behavioural changes which may be attributable to the underlying characteristics of the population and economic environment, such as unemployment rate effects. In order to isolate the effect of the pilot project, complex econometric techniques are used. To ensure the results are valid, the report first applied simple techniques, and then progressively used more precise methods. The final reported results are determined using difference-in-difference methodologies. While the analysis includes individuals from all economic regions, two sensitivity tests are also included - one which calculates the results based on regions with unemployment rates between 6 percent and 13 percent and another that uses regions with unemployment rates between 7 percent and 14 percent.

6.1 Behavioural Response of those Targeted by the Pilot Project

The intent of the pilot project is to increase benefit entitlement for those individuals who experience a gap between the end of their EI entitlement and their next job. To examine this, this evaluation looks at the change in the weeks of benefits, insured hours of work and weeks of work for gappers and exhaustees for those living in the pilot regions. This change is compared to changes that occurred outside the pilot regions, to isolate the impact of the pilot project.

6.1.1 Benefit Weeks Received ^{Footnote 34}

Controlling for various other factors, seasonal gappers received an extra 4.4 weeks of benefits as a result of the pilot during the first year after it's implementation, while all gappers, including seasonal ones, received almost the entire extra five weeks of benefits over this same time period. ^{Footnote 35} This figure remained fairly constant over the three years ending in May 2007, as gappers appear to receive an extra 4.5 weeks as a result of the pilot project over this entire time frame. ^{Footnote 36}

After restricting the economic regions to those with 6 percent to 13 percent unemployment rates or 7 percent to 14 percent unemployment rates, gappers received an additional 4.9 benefit weeks as a result of the pilot, ^{Footnote 37} This is likely a more reliable result, since the economic regions are now closer in characteristics than before.

Evidence suggests that those who exhaust their claims also received extra benefit weeks ^{Footnote 38} ,but this is not surprising since in order to exhaust their claim they have to use all of the benefits they are entitled to. Exhaustees received 4.9 extra weeks of benefits as a result of the pilot during its first year. Similarly, seasonal exhaustees received 4.5 extra weeks of benefits over the same time period.

Limiting the analysis to regions with unemployment rates between 6 percent and 13 percent finds that exhaustees received between 4.6 and 5.0 extra weeks of benefits. Similarly, when the regions are limited to those with unemployment rates between 7 percent and 14 percent the results suggest that exhaustees received between 4.9 and 5.0 extra weeks of benefits. ^{Footnote 39}

6.1.2 Weeks of Work

Weeks of work refers to the number of weeks in which a claimant works at least one insured hour during the EI qualifying period. It is important to note that it is insured hours that are used to determine eligibility and entitlement, not weeks of work. ^{Footnote 40} However, any change in the number of weeks worked as a result of the pilot project is still an important issue to consider as this would constitute a behavioural change.

In the first year after the implementation of the pilot project, there appeared to be no statistically significant impact on the number of weeks worked during the EI qualifying period by seasonal gappers, while in the second year of the pilot, there was a clear behavioural impact, as the number of weeks worked fell by 2.7 weeks as a result of the pilot. ^{Footnote 41} With respect to all gappers, weeks worked appeared to increase in the first year of the pilot by less than a week, but fell by 1.8 weeks in the second year. ^{Footnote 42} These results could be correlated with the fact that claimants now require fewer insured hours to qualify for the same duration of benefits they would have been entitled to had the pilot not been in place. However, it is also possible that employer decisions led to a reduction in work weeks for employees. The change in weeks worked from year to year, especially as it relates to seasonal gappers, could be evidence that claimants and employers began to understand the rules of the pilot project as time went on, and were able to change their working habits to accommodate them.

Limiting the analysis to those EI economic regions with unemployment rates between 6 percent and 13 percent, as well as those with unemployment rates between 7 percent and 14 percent, the results did not change very much for gappers. Considering each unemployment rate band, the results suggest that gappers worked 2.9 and 3.2 fewer weeks over the entire post pilot-implementation period as a result of the pilot respectively. ^{Footnote 43}

Exhaustees did not appear as strongly affected in terms of weeks worked as gappers appeared to be. ^{Footnote 44}   Seasonal exhaustees appeared to work a little more than half a week more during the first year of the pilot than they did before the pilot, whereas the pilot appeared to have no effect on their weeks worked in the second year. All exhaustees on the other hand did not appear to alter their weeks of work during the first year of the pilot, but then worked 0.8 weeks less during the second year.

None of the results are statistically significant when limiting the analysis to regions with unemployment rates that are close together. ^{Footnote 45}

6.1.3 Insured Hours ^{Footnote 46}

In addition to considering changes in the weeks of work, it is also important to determine if there is a behavioural impact with respect to insured hours worked during the EI qualifying period, since it is hours that are used to determine eligibility and the number of EI entitlement weeks a claimant is eligible to receive. Seasonal gappers making claims in the first year of the pilot worked 40.5 fewer hours during the EI qualifying period than they would have worked in the absence of the pilot project, while in the second year of the pilot they worked 59.3 hours less during the qualifying period. Interestingly, when considering all gappers, insured hours increased by 19.2 hours in the first year, but the pilot had no statistically significant impact on hours worked during the second year. However, considering the first three years of the pilot together, it appears that hours worked by all gappers fell by 200.4 hours during the EI qualifying period as a result of the pilot project. ^{Footnote 47} This is a very important result since it suggests that during the third year of the pilot both gappers and employers had a strong reaction to the pilot project in terms of hours of work. This decline may be a function of both workers choosing to work less and employers choosing to provide fewer hours of work to employees. In many cases, working 200 hours less during the EI qualifying period results in a three-to-four week drop in EI entitlement weeks. It is also important to note however, that some individuals make claims on an annual basis and with the additional weeks of benefits, there would have been fewer weeks to work between claims in the second year and third year of the pilot. This may also explain some of the decrease in hours of work, particularly in the latter years of the pilot. The conclusion that could be drawn from all of this is that claimants observe a reduction in hours worked, yet are still receiving more benefits than they would have had if they worked the extra 200 hours and the pilot project was not in place.

When only comparing those individuals in areas with unemployment rates between 6 percent and 13 percent, it appears that gappers tended to work 198.4 fewer hours. By changing the unemployment rate band to between 7 percent and 14 percent for analysis, the pilot project appears to have induced gappers to work 173.1 fewer hours during the EI qualifying period. ^{Footnote 48}

Exhaustees tended to work more hours during the EI qualifying period as a result of the pilot, but the effect is small. Exhaustees worked an extra 17.4 extra hours during the first year of the pilot, while they worked 14.5 more hours during the pilot project's second year. On the other hand, when limiting the analysis to seasonal exhaustees, the results suggest that they reduced their number of hours worked as a result of the pilot. During the first year, seasonal exhaustees worked 23.7 hours fewer hours in the EI qualifying period. In the second year the number of hours worked fell by even more during, with hours worked falling by 43.7 hours.

Results for calculations based on limiting the regions to those within certain unemployment rate bands are not reported since the results are statistically insignificant.

6.2 Behavioural Response of those not targeted by the Pilot Project

Although the main intent of the pilot project is to help gappers, all claimants are able to take advantage of the additional weeks of benefits. This means that the behaviour non-gappers may change as well. It is important to know the extent to which non-targeted individuals receive additional benefits as well as alter their work behaviour, since all of these can be costly to the economy.

6.2.1 Benefit Weeks Received

Evidence suggests that during the first year of the pilot project, non-gappers received an additional 1.8 weeks of benefits. ^{Footnote 49}   This did not change over the first three years of the pilot, as non-gappers received 1.9 extra weeks of benefits as a result of the pilot during this period. ^{Footnote 50} Similarly, non-exhaustees received an extra 1.9 weeks of benefit during the first year of the pilot and received an extra 2.9 benefit weeks during the five-month period from June to October 2005. ^{Footnote 51} These results show that not only are the gappers and exhaustees benefiting from the pilot project, but so are non-gappers and non-exhaustees, some of which could have been gappers or exhaustees had the pilot not been in place.

When limiting the analysis to regions with unemployment rates between 6 percent and 13 percent over the three years after the implementation of the pilot, the results suggest that non-gappers received an additional 1.6 weeks of benefits as a result of the pilot, while limiting the regions to those with unemployment rates between 7 percent and 14 percent suggests that they received 1.7 extra weeks of benefits as a result of the pilot. ^{Footnote 52}

Controlling for various other factors, non-exhaustees receive between 1.9 and 2.9 additional weeks of benefits as a result of the pilot project. ^{Footnote 53} When considering only those non-exhaustees who live in regions where the unemployment rate is between 6 percent and 13 percent the number of weeks of benefits received increased by about 1.4 to 2.1 weeks. ^{Footnote 54} This is similar to the result derived when non-exhaustees are limited to those living in regions with jobless rates of 7 percent to 14 percent. The fact that some claimants used additional weeks of benefits might be associated with them using this time to find a better job match.

6.2.2 Weeks of Work

Non-gappers appear to have reduced the number of weeks worked during the EI qualifying period as a result of the pilot project. Non-gappers worked 0.6 weeks less during the first year of the pilot.  During the second year their number of weeks worked during the EI qualifying period fell even further, by 2.7 weeks relative to the pre-pilot period. ^{Footnote 55}   When considering the first three years of the pilot together, non-gappers worked 1.2 weeks less as a result of the pilot. ^{Footnote 56} This could be the result of a combination of factors. Workers could be realizing that they require fewer hours of work to qualify for the same level of benefits than before the pilot project was initiated, while at the same time employers may alter their behaviour by changing the availability of work.

Only considering non-gappers in regions with unemployment rates between 6 percent and 13 percent, the number of weeks worked falls about 0.5 weeks as a result of the pilot. ^{Footnote 57} Similarly, when only considering regions with unemployment rates between 7 percent and 14 percent, the average number of weeks worked fell by 1.6 weeks. ^{Footnote 58}

Non-exhaustees also appear to have reduced their number of weeks worked prior to implementing a claim. The pilot appears to have induced non-exhaustees to work 1.1 and 1.9 fewer weeks prior to the claim during the first and second year respectively. ^{Footnote 59}

These results are similar to those derived when only comparing regions with similar unemployment rates. When looking at non-exhaustees in regions with unemployment rates between 6 percent and 13 percent, it appears that they worked between 0.9 and 1.3 fewer weeks, while looking at those individuals in regions where the unemployment rate is 7 percent to 14 percent, non-exhaustees appear to work 0.9 to 1.5 fewer weeks as a result of the pilot. ^{Footnote 60}

6.2.3 Insured Hours

As with gappers, there is a concern that the 5-week seasonal pilot project will encourage non-gappers to reduce the number of hours they work during the EI qualifying period, since it is now possible that they will be eligible for the same level of benefits even though they work less. 

In the first year of the pilot, non-gappers worked 13.7 fewer hours, while in the second year, they worked 25.6 fewer hours during the qualifying period. ^{Footnote 61} As was the case with gappers, when considering the first three years of the pilot as a whole, non-gappers worked 150.4 hours less as a result of the pilot project. ^{Footnote 62} While non-gappers also reduced their hours worked in the face of the pilot, they did not reduce them as much as gappers.

When limiting the analysis to non-gappers in regions with similar unemployment rates, the general results did not change. Non-gappers in regions with an unemployment rate between 6 percent and 13 percent worked 113.5 fewer hours during the EI qualifying period, while those in regions with unemployment rates between 7 percent and 14 percent worked 63.9 fewer hours. ^{Footnote 63}

Non-exhaustees also appeared to work fewer hours during the qualifying period. Overall, non-exhaustees worked 37.0 and 55.0 less insured hours during the first and second year of the pilot respectively. ^{Footnote 64} Again, this may have occurred because of the interactions between employees and employers, with employees recognizing they are able to qualify for the same level of benefits with even fewer hours of work and employers offering a different amount of work hours to employees. In this case however, because the reduction in hours worked is so small, it is possible that these individuals were eligible for more benefit weeks than they would have been had they worked the extra 37 to 55 hours in the absence of the pilot project.

The results generated by limiting the non-exhaustees to those in regions with similar unemployment rates are very close to the results produced when considering the entire population. ^{Footnote 65}

6.2.4 Summary

Given the definition of a gapper and an exhaustee, it would be expected that the pilot project would result in everyone from both groups receiving the entire five weeks of extra benefits. The reason this is not the case is because some individuals are already entitled to 41 or more weeks of benefits, limiting them to fewer than five weeks of extra benefits. Nevertheless, the goal of the pilot has been achieved. Seasonal gappers (as well as other gappers) and seasonal exhaustees (as well as other exhaustees) received upwards of four to five additional weeks of benefits, even after accounting for changes in other non-pilot regions, industry, gender, age of the claimant and the regional unemployment rate. However, non-gappers and non-exhaustees also received up to an additional 3.6 extra weeks of benefits.

One of the concerns of this pilot project is that it would reduce the weeks and hours a person would work during the EI qualifying period since they would now be eligible for the same or more number of benefit weeks while working for a shorter period of time. The results seem to suggest that while the average number of weeks and hours worked fell as a result of the pilot project, they did not fall enough to completely cancel out the extra weeks of benefits, meaning that claimants still received more benefits on average. Also important to note is that it is not possible to derive how much of the change in hours of work is a direct result of employee behavioural changes or employer changes.

Still, there are interesting comparisons between the targeted groups and the non-targeted groups with respect to weeks and hours worked. It appears that gappers reduced their weeks and hours of work more than non-gappers. On the other hand, it appears that non-exhaustees reduced their weeks and hours of work more than exhaustees.

6.3 Impact on Job Search Behaviour ^{Footnote 66}

Increasing the number of benefit weeks a person is entitled to decreases the individual's cost of being unemployed. With this cost being lowered, there is less incentive for those currently unemployed to engage in job searching. It is possible that the additional five weeks of benefits reduces the amount of time spent job searching as well as how the individual searches. For example, a person receiving more weeks of EI benefits may engage in more passive forms of job search rather than actively seeking out new employment. It is therefore important to know if the 5-week seasonal pilot project has changed the job search behaviour of the unemployed.

Evidence suggests that the pilot project had no discernible impact on the hours spent ^{Footnote 67} or method of job search of claimants. ^{Footnote 68} The results however, suggest that individuals are less likely to search for part-time jobs when the extra five weeks are available to them. ^{Footnote 69} While this is true for all individuals as a whole, it is also true for those workers who expect to return to the same employer and standard workers. ^{Footnote 70} In addition, it appears that individuals were more likely to move as a result of the pilot project, but this effect declined as the unemployment rate increased. ^{Footnote 71}

7. Response by Employers

Employers are affected considerably by the EI system. In particular, sudden increases in generosity are often followed by concerns of business advocates that employers will have difficulties in adjusting to the new rules as well as the impact of EI changes on employee behaviour. On the other hand, some employee advocates question whether the employer hiring and firing practices may be affected by the new EI rules.

Consequently, two sets of interviews were conducted with seasonal employers, unions and officers in Service Canada offices. The first study ^{Footnote 72} was a set of 33 in-depth one-on-one interviews with senior officials of various seasonal employers, along with three union representatives and four Service Canada representatives. These key informant interviews were conducted in October 2005 and revealed interesting insights into the hiring process. The second follow-up study ^{Footnote 73} was conducted in November-December 2006. In this study, the same questions were asked, but the interviews were conducted by phone to cover a larger number of seasonal employers across Canada. There were 124 interviews conducted in the second set of interviews; 114 representatives from seasonal employers, five union representatives and five HRSDC representatives.

In the first set of interviews, there appeared to be little awareness of the pilot project. After the pilot project had been in place for two years, some employers stated they had heard of the pilot project, suggesting that awareness of the pilot had increased, but still remained low overall. According to the second study, seasonal employers within Ontario and Quebec seemed to be more likely to know about the pilot project, particularly in the construction, tourism and manufacturing sectors.

Most employers interviewed in the first set of interviews rehire seasonal workers, while three-quarters of employers interviewed in the second set of interviews rehire seasonal workers. Initially, most employers stated that the 5-week pilot project did not affect the re-hiring of desired workers, but in the second set of interviews, half of those interviewed thought that the pilot project helped firms rehire seasonal workers while one in seven thought that the pilot caused difficulties in this area.

The employers are usually able to find people to work when they have the need to hire. In particular, hiring employees for short-term, one or two day jobs did not seem to be a problem. Companies usually hire short-term workers based on previous hires and word of mouth. However, a minority of survey respondents, (about one in seven) see the extended benefits from the pilot project as likely to cause some difficulties in hiring employees, with a possible reason being that EI benefits are higher than what the individual could earn working. ^{Footnote 74} Another reason could be that employees derive a greater benefit from the extra free time they have and the EI benefits they collect than they would receive from working full-time. Service Canada and union representatives were more positive about the extension of five weeks than were the employers. Some employers attribute the difficulties in hiring people to the EI system as a whole, but not specifically to the pilot project itself. Most employers encountering problems hiring seasonal workers indicate that skills shortage is the core issue. Companies are generally willing to invest in training their workers with the few unwilling firms saying they have no need for the training.

8. Program Costs ^{Footnote 75}

For the initial pilot, it was estimated that 100,000 claimants would receive an additional $1,000 per claim and thus cost an estimated $100M per year. ^{Footnote 76} Pilot Project #10 was expected to cost $15M less per year as a result of three EI economic regions being dropped from the pilot. ^{Footnote 77}

Two different types of estimates are developed in this section. The first estimate assumes there is no behavioural change as a result of the pilot. That is, it is assumed that individuals did not alter their working behaviour and they would have worked the same number of hours in the EI qualifying period had the pilot project not been in place. In other words, any change in the number of benefit weeks a person is entitled to is directly attributable to the increased generosity of the EI system.

The second set of estimates will attempt to incorporate a behavioural change into the calculations. The change in the number of benefit weeks received will account for other labour market factors that could influence the number of benefit weeks a person receives and only report the number of benefit weeks that are directly attributable to the implement­ation of the pilot project.

8.1 Costs of the Pilot Project without a Behavioural Change

Table A11 demonstrates the cost of the pilot between June 2004 and December 2007 assuming no behavioural change. According to the sample chosen for this analysis, in the first year (June 2004 – May 2005) after the implementation of the pilot, the additional cost per claim was over $1,000.00 for seasonal and non-seasonal gappers and less than $400.00 for both seasonal and non-seasonal non-gappers. Similarly, in the second year (June 2005 – May 2006), the additional cost per claim was over $1,200.00 for gappers and under $400.00 for non-gappers. In the third year of the pilot (June 2006 - May 2007) the average extra benefit received per claim was over $1,200.00 for gappers and again, less than $400.00 for non-gappers. However, while the average cost per claim is lower than expected when looking at all claimants together, the number of claims made in the pilot regions was much higher than the 100,000 previously expected to benefit. Extrapolating these figures with the total number of claims made in the pilot regions shows that the average cost per year of the pilot was $160.8M, $192.8M and $168.4M in the first, second and third year respectively. Over the life of the pilot (June 2004 to December 2007), 2.3 percent of the total cost is received by claimants who are still seasonal gappers after the pilot is introduced, 10.7 percent goes towards gappers who are not seasonal, 12.3 percent goes to seasonal non-gappers and the remaining 74.8 percent is received by non-seasonal non-gappers. Put another way, 14.6 percent of the total expenditure is received by seasonal claimants while 85.4 percent is received by non-seasonal claimants.

8.2 Costs of the Pilot Project including a Behavioural Change

To generate an estimate of the costs that incorporates a behavioural change, some assumptions are made. The number of extra benefit weeks received is taken from Table A6. The figures are based on an average of the first three years after the implementation of the pilot (from June 2004 to May 2007), so it will be assumed that the number of extra weeks of benefits do not change from year-to-year.

According to HRSDC (2008), the average extra benefit weeks received by gappers was 4.53 weeks, while it was 1.88 weeks for non-gappers. Table A12 demonstrates that in the first year of the pilot, the extra cost per claim for gappers was $1,198.96, while the extra cost for non-gappers was $563.47 per claim. Extrapolating these costs to the total number of claims in the pilot regions results in a total cost of $273.4M of which $28.6M (or 10.5 percent) was received by gappers, with the remainder being received by non-gappers. In the second year of the pilot (June 2005 to May 2006), there were fewer regular EI claims, yet the cost of the pilot increased. The cost of the pilot was $283.5M in the second year, with gappers receiving about $31.7M, or 11.2 percent of the total. In the third year, there was a reduction in the number of claims made in the pilot regions. This can almost be entirely explained by the fact that three EI economic regions were dropped from the pilot at this time. The results suggest that the cost of the pilot fell to $257.8M, with gappers receiving 9.1 percent, or $23.4M of the total and non-gappers receiving the rest.

If the pilot project did induce a behavioural change, there are other costs that must be considered both inside the EI program and outside. If the number of hours worked by claimants is reduced as a result of the pilot, then the EI program is forgoing EI premiums for each hour an individual does not work. ^{Footnote 78}

Outside the EI program, as employees work fewer hours there is going to be a slight drop in GDP. HRSDC (2007) estimated that while claimants worked more hours during the first year as a result of the pilot, gappers and non-gappers reduced their hours worked during the EI qualifying period in the second year.  Estimates suggest that hours worked might have fallen by more than 5 million hours in the first year and by 11 million hours in the second year. ^{Footnote 79}

8.3 Summary

The findings above suggest that the costs of the 5-week seasonal pilot project are higher than what was estimated when the pilot project was being designed. It would also appear that the costs taking into account a behavioural change are much higher than those ignoring the behavioural impact. Assuming no behavioural change, the estimates of actual costs paid out by the program are between $160.8 million and $192.8 million per year over the first three years, while incorporating a behavioural change inflates the costs by even more, as the annual cost in each of the first three years of the pilot is estimated to be between $257.8 million and $283.5 million. This does not include the value of the premiums that the EI program no longer receives because of the reduction in insured hours worked by claimants. Table 1 below shows a comparison of costs using the initial pre-pilot estimates, the calculations assuming no behavioural change and then including a behavioural change over the first three years of the pilot. ^{Footnote 80}

• Reference from the above tableNote : Figures may not add perfectly due to rounding.
  Calculated based on a 20% random sample of claimants who made at least one claim consisting entirely of regular benefits between June 2003 and December 2007.
  Years begin in June because the pilot project began in June of 2004.

In addition to the monetary costs, there are other costs which cannot be as easily quantified. For example, the economic costs associated with a reduction in hours worked cannot be easily quantified monetarily, but are still a significant cost of this pilot project. The results suggest that total hours worked fell by more than 5 million hours in the first year of the pilot and fell by an additional 11 million hours in the second year. Clearly, the economic costs of this pilot project are much higher than just the monetary costs of increasing EI benefit by up to five weeks.

9. Conclusions and Recommendations

The pilot project on increased weeks of EI benefits (Pilot Project #6) was implemented in 24 high unemployment regions across Canada on June 6, 2004 and ended on June 4, 2006. This pilot was then extended in 21 of the original 24 high unemployment regions through the extended EI pilot project (Pilot Project #10) which started on June 11, 2006 and ended on February 28, 2009. Together, these projects are collectively known as the 5-week seasonal pilot project. The 5-week seasonal pilot project increased the maximum number of benefit weeks a claimant could receive by up to five weeks with the maximum number of benefit weeks remaining at 45 weeks.

The rationale for this project was based on the assumption that seasonal workers experiencing income gaps are undergoing considerable hardship. There is significant evidence that suggests that this assumption does not always hold. There is evidence that seasonal workers are heterogeneously spread out over the workforce and that they are as able to smooth their consumption over time as other individuals. Overall, these findings would suggest that seasonal workers and are not unique in their difficulties of establishing a regular income stream.

Overall, the 5-week seasonal pilot project was well targeted in that gappers and exhaustees were entitled to the extra weeks of benefits and the majority of these individuals used the extra weeks. While non-gappers and non-exhaustees also made use of the program, it is important to remember that because of how claimants are labelled as a gapper, non-gapper, exhaustee or non-exhaustee, it is impossible to know if a non-gapper would have been a gapper or if a non-exhaustee would have been an exhaustee had they not been eligible for the extra five weeks of benefits. Nevertheless, it appears as though those who are labelled as non-gappers and non-exhaustees also received extra benefit weeks, but not as many as gappers and exhaustees received. However, because there are more non-gappers and non-exhaustees than gappers and exhaustees, the extra benefit weeks received by the non-gappers and non-exhaustees accounts for the majority of the costs related to the pilot project. Cost estimates suggest that the monetary cost of the pilot project was higher than the $100 million per year originally estimated at the outset of the pilot and is between one-and-a-half and three times this amount. The true economic cost of this pilot was even higher than that, as there was a reduction in the hours worked by claimants leading up to an EI claim.

In addition to more benefit weeks being paid out to claimants, there were other behavioural changes occurring that cannot be as easily quantified in terms of their cost. The evidence suggests that claimants reduced the number of weeks and hours they work in the EI qualifying period as a result of the pilot. This could be the result of workers choosing to work fewer hours and employers changing the amount of work they offered employees. In general, all types of claimants worked fewer weeks as a result of the pilot, with the only possible exception being exhaustees, as they did not change their weeks worked. Insured hours appeared to have fallen as well, with all claimants experiencing a reduction in the number of hours worked during the qualifying period.

In addition to employment behaviour, the effect of the pilot project on job search behaviour has also been considered. In general, it appears that job search behaviour was not altered as a result of the pilot. However, there is evidence that job seekers were less likely to search for part-time jobs when the extra five weeks were available to them.

Employers' reactions to the additional weeks of benefits available to EI claimants were also observed. There were two sets of interviews conducted with union officials, Service Canada representatives and officials within seasonal employers. During the initial set of interviews, the majority of employers said that the pilot project did not effect the re-hiring of desired workers, while in the second set of interviews half of employers thought that the pilot project helped firms rehire seasonal workers. At the same time, qualitative analysis showed that some firms felt that the pilot caused trouble in this area by offering individuals more money than they could earn working.

Since the pilot project was implemented for the purpose of collecting information, no specific recommendations are being made at this time.

Annex A:  Claimant Tables

• Reference from the above tableNote: Calculated based on a 20% random sample of claimants who made at least one claim consisting entirely of regular benefits between June 2003 and December 2007. Multiplying the number of sample claims by five will provide an estimate of the claimant population for that year. This number may differ from the MAR, since the MAR considers anyone who received at least one week of regular benefits during the course of their claim as a regular claimant even if they also received some other type of benefit over the course of their claim as well.
  Years begin in June because the pilot project began in June of 2004.

• Reference from the above tableNote: Calculated based on a 20% random sample of claimants who made at least one claim consisting entirely of regular benefits between June 2003 and December 2007. Multiplying the number of sample claims by five will provide an estimate of the claimant population for this time period. This number may differ from the MAR, since the MAR considers anyone who received at least one week of regular benefits during the course of their claim as a regular claimant even if they received some other type of benefit over the course of their claim as well.
  Results are only presented to December 2007, as this is the most recent period where results are reliable.

• Reference from the above tableNote: ***significant at a 1% level ** significant at a 5% level *significant at a 10% level
  Calculated based on a 10% random sample of claimants who made at least one claim consisting entirely of regular benefits between June 2003 and December 2007.
  Years begin in June because the pilot project began in June of 2004.
  Results are only presented to October 2005 because this was the most recent period where results were reliable at the time of analysis.
  ¹These are highlights from a technical report. Full results and explanation of the methodology can be found in Human Resources and Social Development, September 2007.

• Reference from the above tableNote: ***significant at a 1% level ** significant at a 5% level *significant at a 10% level
  Calculated based on a 10% random sample of claimants who made at least one claim consisting entirely of regular benefits between June 2003 and December 2007.
  Years begin in June because the pilot project began in June of 2004.
  Results are only presented to October 2005 because this was the most recent period where results were reliable at the time of analysis.
  ¹These are highlights from a technical report. Full results and explanation of the methodology can be found in Human Resources and Social Development, September 2007.

• Reference from the above tableNote: ***significant at a 1% level ** significant at a 5% level *significant at a 10% level
  Calculated based on a 10% random sample of claimants who made at least one claim consisting entirely of regular benefits between June 2003 and December 2007.
  Years begin in June because the pilot project began in June of 2004.
  Results are only presented to October 2005 because this was the most recent period where results were reliable at the time of analysis.
  ¹These are highlights from a technical report. Full results and explanation of the methodology can be found in Human Resources and Social Development, September 2007.

• Reference from the above tableNote: ***significant at a 1% level ** significant at a 5% level *significant at a 10% level
  Calculated based on a 20% random sample of claimants who made at least one claim consisting entirely of regular benefits between June 2003 and December 2007.
  Only data up to May 2007 was used, as this was most recent period where results were reliable at the time of analysis.
  ¹These are highlights from a technical report. Full results and description of the methodology can be found in Human Resources and Social Development Canada, 2008.

• Reference from the above tableNote: *** indicates significant at 99% confidence level, ** 95%, and * 90%.
  All regressions include controls for gender, age, household size, family type, education, regional unemployment rate, industry of ROE job, personal income was greater then 50% of household income and whether the person moved. Figures in parentheses are standard errors.
  Data: Canadian Out-of-Employment Panel (COEP) Survey, Cohorts Apr.-Jun. 2003 & Oct.-Dec. 2004.
  "Do Not Expect to Return" includes respondents who do not expect to return to the same employer, while "Expect to Return" includes respondents who expect to return to the same employer. "Non-Standard Worker" includes respondents who were seasonal, part-time or temporary workers at the time of their job loss, while "Standard Worker" includes all other respondents.

• Reference from the above tableNote: Data: Canadian Out-of-Employment Panel (COEP) Survey, Cohorts Apr.-Jun. 2003 & Oct.-Dec. 2004.
  Type of job search used was self reported. Figures represent the proportion of the sample who utilized the particular job search method. Control group is the non-pilot regions and the treatment group is the pilot regions.
  ¹These are highlights from a technical report. Full results and description of the methodology can be found in Stewart, 2007.

• Reference from the above tableNote: *** indicates significant at 99% confidence level, ** 95%, and * 90%.
  All regressions include controls for gender, age, household size, family type, education, regional unemployment rate, industry of ROE job, personal income was greater then 50% of household income and whether the person moved. Figures in parentheses are standard errors.
  Data: Canadian Out-of-Employment Panel (COEP) Survey, Cohorts Apr.-Jun. 2003 & Oct.-Dec. 2004.
  "Do Not Expect to Return" includes respondents who do not expect to return to the same employer, while "Expect to Return" includes respondents who expect to return to the same employer. "Non-Standard Worker" includes respondents who were seasonal, part-time or temporary workers at the time of their job loss, while "Standard Worker" includes all other respondents.
  ¹These are highlights from a technical report. Full results and description of the methodology can be found in Stewart, 2007.

• Reference from the above tableNote: *** indicates significant at 99% confidence level, ** 95%, and * 90%.
  All regressions include controls for gender, age, household size, family type, education, regional unemployment rate, industry of ROE job, personal income was greater then 50% of household income and whether the person moved. Figures in parentheses are standard errors.
  Data: Canadian Out-of-Employment Panel (COEP) Survey, Cohorts Apr.-Jun. 2003 & Oct.-Dec. 2004.
  "Do Not Expect to Return" includes respondents who do not expect to return to the same employer, while "Expect to Return" includes respondents who expect to return to the same employer. "Non-Standard Worker" includes respondents who were seasonal, part-time or temporary workers at the time of their job loss, while "Standard Worker" includes all other respondents.
  ¹These are highlights from a technical report. Full results and description of the methodology can be found in Stewart, 2007.

• Reference from the above tableNote: Note that numbers may not add perfectly due to rounding. Results are only presented to Dec. 2007, as this is the most recent period where results are reliable. The # of sample claims is based on a 20% random sample of claimants who made a pure regular claim during this time period. The # of total claims is taken from the Status Vector header and represents the number of pure regular claims initiated over the study period. The analysis here assumes that the claims selected as part of the sample are representative of the total claims made over this period. This analysis may underestimate the true cost of the pilot project, as each seasonal claimant receives more extra benefits than other claimants and is going to have many claims over the study period while other types of claimants may only have one claim over the entire period.

• Reference from the above tableNote: Note that numbers may not add perfectly due to rounding. Results are only presented to Dec. 2007, as this is the most recent period where results are reliable. The # of sample claims is based on a 20% random sample of claimants who made a pure regular claim during this time period. The # of total claims is taken from the Status Vector header and represents the number of pure regular claims initiated over the study period. The analysis here assumes that the claims selected as part of the sample are representative of the total claims made over this period. This analysis may underestimate the true cost of the pilot project, as each seasonal claimant receives more extra benefits than other claimants and is going to have many claims over the study period while other types of claimants may only have one claim over the entire period.

Annex B:  Evaluation Questions

The following section outlines the focus and analytical requirements for a credible evaluation of the pilot project to extend EI benefits up to a maximum of five weeks in high-unemployment regions. It delineates and lists clearly and precisely the proposed evaluation issues to be addressed, the result indicators, and the proposed methodologies to collect and analyse the data.

The proposed evaluation issues are divided into the following broad categories:

• The Rationale for the Pilot Project
• Objectives and Achievements
• Program Impact and Effects 

Q-1: What are the factors underlying the EI support for seasonal workers and part-time or non-permanent workers?

Q-2: Do the new design features of the pilot project initiative support this goal?

Q-3: Are the five weeks of extra EI benefit entitlement shortening the incidence and duration of 'gaps'?

Q-4: To what extent does the existence of EI gaps create economic difficulties for seasonal workers and part-time or non-permanent workers?

Q-5: Are the extra five weeks of EI benefit entitlement leading to longer periods of subsequent employment?

Q-6: Are the extra five weeks of EI benefit entitlement leading to shorter durations of subsequent unemployment spells?

Q-7: Are the extra five weeks of EI benefit entitlement encouraging greater labour force participation?

Q-8: Are the extra five weeks of EI benefits changing the job search behaviour of workers?

Q-9: Are the extra five weeks of EI benefits affecting the working on claim behaviour of EI claimants?

Q-10: Are the five weeks of EI benefits affecting the hiring and lay-off behaviours of employers?

Q-11: What are the unintended consequences and impacts of extending EI Benefit Entitlement by Five Weeks for EI Gappers?

Q-12: What are the budgetary costs of the program?

Annex C:  List of Reports Used in Evaluation

COMPAS Inc., Second Study among Key Informant for an Evaluation of the Pilot Project of the 5-Week Extension of Employment Insurance Benefits, Prepared for Human Resources and Social Development Canada, January 2007.

Gomez, Rafael & Morley Gunderson, Employment Patterns of Non-Standard Workers: Analysis Using 2001 GSS. Prepared for Human Resources Development Canada, 2005.

GPC Research, Key Informant Interviews for the Two year Pilot Project to Extend Employment Benefits by Five Weeks , Prepared for Human Resources and Skills Development Canada, 2006.

Gray, David, Rick Audas, Ted McDonald and Thomas Lang, Repeat Use and the Persistence of UI/EI Receipt in Canada, July 2005.

Gunderson, Morley., Review of Research and Policy Literature on the Impacts of Employment Insurance Programs for Seasonal Workers, Report to HRSD. Revised March 2006.

HRSDC, Behavioural Impacts of the Pilot Project on Increased Five Weeks of EI Benefits, Evaluation Directorate, December 2008.

Human Resources and Skills Development Canada, Employment Insurance: 2004 Monitoring and Assessment Report, Employment Programs Policy and Design, Employment Insurance Policy, 2005.

Human Resources and Skills Development Canada, Technical Note on the Change in Hours of Work as a Result of the Five-Week Seasonal Pilot Project, Strategic Evaluation, Evaluation Directorate, Strategic Policy and Research, September 2009.

Human Resources and Social Development Canada, Behavioural Impacts of the Pilot Project on Increased Five Weeks of EI Benefits Based on Difference-in-Difference and Econometrics Analyses, Employment Insurance Evaluation, Policy Evaluation, Evaluation Services, September 2007.

Human Resources and Social Development Canada, Data Collection and Analysis: Evaluation of the 5-Week Extended EI Benefits Pilot Project, Audit & Evaluation Branch, March 2006.

Human Resources and Social Development Canada, Summative Evaluation of EI Part I: A Summary of Evaluation Knowledge to Date, EI Evaluation, Audit and Evaluation Directorate, Strategic Analysis, Audit and Evaluation Branch, June 2006.

Kapsalis, Costa., Sectoral Impacts of the 5-Week Extension of EI Benefits, Produced for EI Evaluation, Submitted to HRSDC, Employment Insurance Evaluation, 2006.

McDonald, Ted. & David Gray., A Statistical Profile of Seasonal and Non-Seasonal Workers and EI Gappers: Within the framework of the Evaluation of the two-year Pilot Project to Extend Employment Insurance Benefits by Five Weeks, Prepared for Human Resources and Social Development Canada, April, 2006.

Prime Minister's Task Force on Seasonal Work, The Seasonal Economy: Responding with Care, Liberal Task Force Report, November, 2004.

Stewart, Jennifer M., Impact on Job Search Patterns From the Pilot Project on Extending EI Benefit Duration by 5 Weeks, Prepared for Human Resources and Social Development Canada, August 2007.

Annex D:  Supplementary Tables

Eligibility for EI is based on variable entrance requirements (VER). The higher the unemployment rate in a region, the fewer hours a person must work to qualify for benefits. Similarly, the higher the unemployment rate, the more weeks of benefit a person is eligible for, for a given number of hours worked. The Schedule Before Pilot (Table D1) shows the number of weeks a claimant is eligible to receive benefits for (based on the local unemployment rate and hours worked during the qualifying period) under regular EI rules while the Schedule After Pilot (Table D2) shows the number of weeks a claimant is eligible to receive benefits for under the rules of the pilot project.

• Reference from the above tableNote^*The Increased Weeks of EI Benefits (established from June 6, 2004 to June 4, 2006) was replaced by the Extended EI Benefits pilot project on June 11, 2006. The Best 14 Weeks (B14) Pilot Project began on October 30, 2005 and the Increase in Allowable Earnings (WWOC) and Increased Access to EI (NERE) Pilot Projects began on December 11, 2005.

Table D4 Timeline for Pilot Project

June 2001
Three years prior to the start of the pilot project on Increased Weeks of EI Benefits

May 2004
Announcement of pilot project on Increased Weeks of EI Benefits (pilot project #6)

June 6, 2004
Start of pilot project on Increased Weeks of EI Benefits (pilot project #6)

December 5, 2004
Minimum possible benefit termination for someone in a designated region making a claim in the first week of the pilot project (26 entitlement weeks is the minimum number of entitlement weeks for claimants designated regions).

April 17, 2005
Maximum possible benefit termination for someone in a designated region making a claim in the first week of the pilot project (45 weeks after the start of the pilot).

May 2006
Announcement of the Extended EI Benefits Pilot Project

June 4, 2006
End of pilot project on Increased Weeks of EI Benefits (pilot project #6)

June 11, 2006
Start of pilot project on Increased Weeks of EI Benefits (pilot project #10)

April 2006
Maximum possible benefit termination for someone in a designated region making a claim in June 2005, the last week of the first year of pilot project #6. That is, 45 weeks after June 2005.

December 9, 2007
End of pilot project on Increased Weeks of EI Benefits (pilot project #10)
Announcement of the extension of pilot project #10 until June 6, 2009

April 2008
Maximum possible benefit termination for someone in a designated region making a claim in June 2007, the last week of the first year of pilot project #10. That is, 45 weeks after June 2007.

October 2008
Maximum possible benefit termination for someone in a designated region making a claim in the last week of pilot project #10. That is, 45 weeks after December 9, 2007.

June 6, 2009
End of pilot project on Increased Weeks of EI Benefits (pilot project #10)