Seasonal work and Employment Insurance use
Shawn de Raaf, Costa Kapsalis, and Carole Vincent
S easonal work has long been an important aspect of the Canadian labour market. With a large resource sector and one of the most varied climates in the world, Canada naturally exhibits large seasonal fluctuations in output and employment. In recent decades, however, the importance of seasonal work has gradually diminished as industries have modernized and diversified. The average monthly swings in employment due to seasonality declined during the 1976-1997 period (Marshall 1999; Guillemette, L'Italien and Grey 2000 ). 1 Two principal trends have contributed to this reduction: a technology-driven decrease in seasonality within traditionally seasonal industries, and an overall drop in the employment share of these industries as a result of more demand for services and less demand for manufactured goods.
However, seasonal jobs continue to account for a large share of employment in some regions. The Atlantic region in particular remains well above average, mainly because of its highly seasonal industries. Moreover, Canada has more seasonal workers relative to other countries with a similar climate. Among the Nordic countries (Finland, Sweden, Denmark, Iceland, and Norway), where similar seasonal employment trends would be expected, only Finland exhibited greater seasonal fluctuations in employment over the 1994-1998 period than Canada (Grady and Kapsalis forthcoming ).
While seasonal work may be preferable for some, such as students, it may not be the optimal pattern for many workers who from year to year face various barriers to secure, year-round employment. Since their financial resources may be uncertain for large parts of the year, many seasonal workers rely on Employment Insurance (EI) to stabilize their income in the off-season. However, not all workers resort to EI following their layoff, either by choice or because their seasonal employment does not provide them with enough hours of work to qualify.
This article addresses disparities in the measurement of seasonality by proposing definitions that distinguish between seasonal workers and seasonal jobs. Using longitudinal data from the Survey of Labour and Income Dynamics (SLID) for 1993 to 1998, the many dimensions of seasonality are examined to determine the extent to which each contributes to frequent reliance on EI benefits. The article also looks at which characteristics distinguish seasonal workers who frequently rely on EI benefits from those who claim infrequently or not at all. Over the 1993-1998 period, a majority of seasonal workers regularly relied on EI following a seasonal job spell, but almost one-fifth did not claim at all. Also pointed out is the significant variation in the characteristics of seasonal workers according to their reliance on EI.
Using SLID to measure long-term seasonality
Measuring seasonality on an individual basis is challenging since seasonal jobs account for only a small fraction of the millions of hirings and separations that give rise to seasonal employment patterns. It is easier to classify a job rather than a worker as seasonal. By definition, seasonal jobs provide temporary work that is expected to last only until the end of a 'season'—the period for which services are in demand. In contrast, seasonal workers are individuals who face annual spells of unemployment because of regular fluctuations in demand for their particular set of skills and experience. They may work one or more jobs, not all necessarily considered seasonal, in such a way that their annual employment displays a seasonal pattern.
A variety of measures have been used to identify the extent of seasonal work. While some give an indication of the incidence of EI use among seasonal workers or the seasonality of frequent claimants' EI patterns (Table 1 ), they do not directly measure the relationship between seasonal work and frequent EI use. Such analysis requires a longitudinal source, such as SLID, which captures both work and EI use patterns over time. 2
Self-identification raises concerns about the accuracy of respondent perceptions of the seasonal nature of their work. For instance, respondents may incorrectly identify their jobs as seasonal if they work in seasonal employment but the job does not end for seasonal reasons. Or, they may simply not be aware the job ended for seasonal reasons and therefore incorrectly indicate other reasons. As well, a seasonal worker may work a variety of temporary jobs, not all of which end for seasonal reasons.
SLID provides an opportunity to move beyond self-identification and to classify seasonal workers according to employment patterns over several years. By comparing job separations and work absences from year to year, it is possible to identify long-term seasonal workers and then link their seasonal employment patterns to any EI claims.
Long-term seasonal workers are defined as persons aged 18 to 59 in 1993 who were not full-time students at any point during the 1993-1998 period 3 and who had at least three paid-job (or self-employed in fishing) spells ending within the same three-month 'off-season' over the five-year period 1993-1997 or 1994-1998. Jobs could not last more than nine months. With this 'mechanical' definition of seasonal work, 4.4% of all employees and self-employed fishers were seasonal workers over the six-year period.
Comparing measures of long-term seasonal workers
The mechanical definition can be compared with the two common definitions of seasonal work: the self-reported definition, which estimates the number of workers who report they experienced a job loss or absence from work for seasonal reasons, and the industry-based definition, which estimates the number of workers in traditionally seasonal industries. Under the self-reported definition, 2.6% of employees and self-employed fishers reported seasonal reasons for their job spell ending in 1998 (Table 2 ). Under the industry-based definition, the figure was 2.1%.
For consistency with the mechanical definition, these job spells could not have lasted more than nine months. A comparison of these two definitions with the mechanical definition also requires that these workers had a seasonal job spell in at least two of the previous four years (1994 to 1997). The addition of this long-term dimension reduced the incidence of seasonal work from 2.6% to 1.0% under the self-reported definition and from 2.1% to 0.9% under the industry-based definition. The mechanical definition identified the highest percentage of workers in 1998 as seasonal. Overall, 2.6% of employees and self-employed fishers met at least one of the definitions. These estimates are low, likely because seasonal workers needed to have had at least two seasonal jobs spells between 1992 and 1997 in addition to their spell in 1998.
The mechanical definition identified the largest pool of seasonal workers (Table 3 ). It accounted for 85.0% of workers with a job spell that ended in 1998 who satisfied at least one of the three definitions of long-term seasonal workers (2.2 of the 2.6% of employees and self-employed fishers). The self-reported definition provided the next largest estimate, 38.5% of long-term seasonal workers (1.0 of 2.6%), followed by the industry-based definition. Moreover, a significant proportion (41.6%) of seasonal workers satisfied the mechanical definition alone. Self-identification or the industry-based definition failed to capture two-fifths of potential seasonal workers. On the other hand, the mechanical definition excluded 15% of workers who satisfied one of the other definitions. On balance, the mechanical definition of seasonal work appears to be the best in identifying the commonalities shared by workers who face regular seasonal layoffs and must rely on EI benefits, regardless of their industry or perception of the nature of their work.
The link to EI benefits
To determine the relationship between seasonal work interruptions and reliance on EI, SLID respondents were classified as seasonal workers if they had three unemployment spells occurring in the same 'off-season' in either the 1993-1997 or the 1994-1998 period. A job spell was associated with an EI spell if the individual received EI benefits within three months following the end of the job spell (Table 4 ). 4
More than one-sixth (17.3%) of long-term seasonal workers did not receive EI benefits following any of their three seasonal job spells. In total, about 61% of seasonal jobs were followed by EI. (The EI Coverage Survey found EI use
to be 61% among workers who identified their last job as seasonal.)
The finding that three-fifths of seasonal job spells led to EI receipt indicates that long-term seasonal workers face significant barriers in finding a new job. Long-term seasonal workers were more likely to be older, male, less educated, living in regions with high unemployment rates, living with a partner, and living in the Atlantic provinces or Quebec (Table 5 ).
Workers who never received EI or received EI only once were nearly evenly divided between men and women, but men accounted for over two-thirds of those claiming two or three times. Seasonal workers relying most frequently on EI tended to be older than other EI users. Among workers who received EI following each seasonal job spell, the proportion 40 and older was nearly double that of those who never claimed or claimed only once (39.3% versus 21.2%). The majority of long-term seasonal workers had not graduated from postsecondary education—less than one-third had postsecondary credentials, compared with 45.2% of workers who never relied on EI.
The use of EI is related to local job opportunities for seasonal workers as well as regional eligibility rules. The EI program has variable entry requirements that fluctuate according to local labour market conditions. A worker living in a region with a lower unemployment rate will not only need more hours of work to qualify for EI, but will also receive fewer weeks of benefits for a given amount of work than someone in a high unemployment region. Nearly half of seasonal workers who never claimed EI lived in low unemployment regions, while over two-thirds with three years of receipt lived in regions with unemployment rates of over 9%.
Long-term seasonal workers who relied most intensively on EI were more likely to live in Atlantic Canada and Quebec, where unemployment rates tend to be higher and seasonal work is more integral to the economy. The majority of workers who claimed EI in only one or two years lived in Ontario or the Western provinces. Although these provinces had a significant population of seasonal workers, because of either stricter regional eligibility requirements or a greater availability of off-season work, the end of a seasonal job did not necessarily lead to an EI claim.
The family circumstances of long-term seasonal workers varied somewhat according to EI use. While the majority of seasonal workers lived with a partner, this share became even larger as reliance on EI increased. However, this does not mean that seasonal workers claiming EI more frequently were better off financially. Although the distribution of family income varied only slightly among the four types of EI claimants, seasonal workers with one or no claims were more likely to be in the highest ($60,000 plus) category, even though they were less likely to be living in a household with another potential adult earner. Workers with two or three claims were more likely to be in the lowest income category (under $35,000).
Workers who claimed EI after only one of their three seasonal job spells appeared to experience less financial hardship than other seasonal workers—even those who never claimed EI at all. They were much more likely to be living in households with higher family incomes. This may reflect their personal circumstances, since they tended to have the highest educational attainment and were least likely to be living in high unemployment regions. These seasonal workers appeared to have greater flexibility in their decision to claim EI, likely because better work opportunities were available to them.
Long-term seasonal workers can be found across Canada, in regions with a diversity of economic conditions. The large percentage of seasonal workers living in regions with relatively low levels of unemployment belies the stereotype of seasonal EI claimants: persons living in regions with poor economic conditions and heavily dependent on traditionally seasonal industries. Nevertheless, seasonal workers relying more frequently on EI tended to live in regions with fewer employment opportunities.
A seasonal worker's economic circumstances and personal characteristics appear to be key factors in determining the degree of reliance on EI (Table 6 ). Individuals who did not receive EI following any of their three seasonal jobs had the highest incidence of potential non-eligibility for benefits because of insufficient hours of work. The incidence of multiple jobholding before the work interruption was also highest among these individuals. Consistent with the higher incidence of multiple jobholding, workers who never claimed EI had the highest incidence of part-time re-employment. However, their incidence of full-time re-employment was the lowest; they were employed full time within three months following 50% of their work interruptions, roughly 10 percentage points lower than those with one or two seasonal spells leading to EI.
Seasonal workers continue to be a large and growing proportion of EI beneficiaries. Despite a general decrease in the proportion of frequent claimants from 1999-2000 to 2000-2001, frequent seasonal claims declined by only 3.7% compared with a 5.6% drop for frequent non-seasonal claims. The relative stability of seasonal claims is "not surprising, as the nature of some seasonal work does not necessarily lead to a decline in claims in periods of strong economic growth" (HRDC 2002 ).
One reason for the growing proportion of seasonal workers among EI claimants may be that the 1996 change from a weeks-based to an hours-based system for determining eligibility has had a positive effect on their EI eligibility and entitlement. The switch was made in part to address concerns that a large and growing proportion of workers were not eligible for EI benefits should they become unemployed. However, it also meant that weeks worked by seasonal workers—who tend to work more hours per week—would now be insured to a greater extent, allowing many to qualify sooner for benefits. Indeed, the reforms resulted in a marginal increase in eligibility and an increase of 1.6 weeks of entitlement among seasonal claimants (HRDC 2001 ).
However, not all seasonal workers were positively affected. Seasonal claimants with less than 30 hours of work per week lost significantly in terms of EI eligibility compared with other claimants (HRDC 2001 ). (They were 21 percentage points less likely to qualify for EI, and those who did qualify received 2.6 weeks less of entitlement.)
Despite the large contribution of seasonal work to EI dependency, not all seasonal workers are frequent EI claimants. While a majority do rely on EI on a regular basis, a significant proportion do not claim at all. They are not necessarily able to avoid relying on EI by doing better in the labour market. They are younger and more likely to live in regions with relatively good employment opportunities; however, at the same time they are more likely to have a lower attachment to the labour market or to be in a precarious employment situation—combining multiple, possibly part-time jobs to provide year-round employment. Conversely, seasonal workers who rely the most on EI face significant barriers to securing non-seasonal employment. They are older, less educated, and live in regions with the poorest employment opportunities.
Introduced in 1993, the Survey of Labour and Income Dynamics (SLID) is well suited to the study of recurrent or long-term seasonal workers since it is designed to track the economic well-being of respondents over time. SLID samples working-age individuals who do not live on reserves or in institutions and who are not serving in the Canadian Forces. Individuals are interviewed over six years, with a new panel of respondents selected every three years. Each panel contains about 15,000 households representing about 30,000 individuals aged 16 and older.
Information is collected in two annual interviews: labour in January and income in May. The labour interview collects such information as the person's employment during the past year, household composition, and educational activity. The income interview collects information on an individual's income and its sources during the previous year. This interview is not necessary if the respondent gives Statistics Canada permission to use tax records. As a result, most respondents do not have to complete the income interview.