The LWF is constructed by integrating data from four sources. First, information on the demographic characteristics and information on certain financial characteristics of individuals is drawn from T1 personal taxation files. Among these variables are sex, age, marital status, and annual net self-employment income. Second, information on the jobs held by individuals is drawn from T4 records. This includes information on earnings and union dues associated with each job. Third, information is drawn from the Record of Employment that employers submit to Human Resources and Skills Development Canada (HRSDC) when an employee working in insurable employment has an interruption in earnings. This includes information on the start date and end date of the job and the reason for job termination. Finally, information is drawn from the Longitudinal Employment Analysis Program of Statistics Canada. This includes information on the business enterprise in which jobs are located, including the estimated size of the business and the industry in which the business is located.
Individuals aged 33 to 38 in 1983 who died within the reference period are excluded.
It is worth noting that labour force participation rates among older Canadians reached a low point between the mid- to late 1990s, declining to 85.7% among men aged 50 to 54 in 1998, to 70.7% among men aged 55 to 59 in 1998, and to 43.4% among men aged 60 to 64 in 1995. The men and women in our sample were aged 46 to 54 over this period, and hence less likely than their older counterparts to leave the labour force during this period.
Spells of unincorporated self-employment (i.e., spells of net self-employment income reported on the T1 tax form) are identified and included in the analysis. However, some measurement issues must be noted. Individuals observed in continuous spells of unincorporated self-employment may have closed one business and opened another and, in this respect, have changed jobs. Such changes cannot be identified and may result in an undercounting of employment spells. Conversely, unincorporated self-employed individuals who subsequently incorporate their business and pay themselves wages and salaries (i.e., T4 earnings) cannot be differentiated from unincorporated self-employed individuals who close their business and take paid employment in another firm. Both cases are counted as two employment spells.
Late filing may also be responsible for some observed gaps.
Fewer than 5% of worker-employer spells are characterized by a break of two years or more when only paid employment is considered. Self-employment income is more frequently characterized by gaps of two years or more in income from that source. Whether this reflects true employment changes or ongoing fluctuations in the self-employment activity cannot be said.
The spike at observed job duration of 28 years represents workers who spent 28 years or more with the same employer, and these job spells are likely both left-hand and right-hand censored in the 28-year LWF panel.
However, readers will recall that women were almost twice as likely as men to be excluded from the sample because they did not have any earnings over the four years from 1983 to 1986.
The 10-year threshold is also used in tables presented in Labour Force Survey Annual Averages.
presents the corresponding distribution when the sample restriction requiring workers to be present in the LWF between 1983 and 1986 is lifted.
This does not include the additional restrictions that Cahill et al. (2006) imposed on their sample. If long-term employment is defined as at least ten years of earnings from the same firm or organization, 74% of our sample is employed in such employment—76% of men and 71% of women.
Layoffs were counted over the period from 1983 to 2008 since, in later years, it is not possible to distinguish with confidence between permanent and temporary layoffs.
The LWF does not currently contain information on sources of income such as tips and commissions.
These estimates are for the time period 2000 to 2008; the study further shows there are no long-term trends in hiring or reallocation rates.
Although jobs are no longer allowed one-year gaps in order to be considered a single job spell, one-year gaps may represent a legitimate extended leave from a job or may be the result of late filing on the part of a company (T4 records filed late would not appear in the Longitudinal Worker File (LWF)). Spells first observed in the data base in 1984 may not have started that year, but rather some time before 1983, the first year of LWF. For that reason, only jobs first observed in 1985 or later are included in the analysis. At the same time, this study may have inaccurately declared some spells as completed, when in fact they were not.